I360 V11.1 Advanced Speech Analytics User Guide

Advanced Speech Analytics
User Guide
Version 11.1
Document Revision 1.06
Confidential and Proprietary Information of Verint Systems Inc.

11/540,185; US 11/540,320; US 11/540,900; US 11/540,902;
© 1992–2011 Verint Systems Inc. All Rights Reserved Worldwide. US 11/567,808; US 11/567,852; US 11/583,381; US
Confidential and Proprietary Information of Verint Systems Inc. 11/608,340; US 11/608,350; US 11/608,358; US 11/616,490;
US 11/621,134; US 11/691,530; US 11/692,983; US
All materials (regardless of form and including, without 11/693,828; US 11/693,923; US 11/693,933; US 11/712,933;
US 11/723,010; US 11/742,733; US 11/752,458; US
limitation, software applications, documentation, and any other
11/776,659; US 11/824,980; US 11/831,250; US 11/831,257;
information relating to Verint Systems, its products or services)
are the exclusive property of Verint Systems Inc. Only expressly US 11/831,260; US 11/831,634; US 11/844,759; US
11/872,575; US 11/924,201; US 11/937,553; US 11/959,650;
authorized individuals under obligations of confidentiality are
US 11/968,428; US 12/015,375; US 12/015,621; US
permitted to review materials in this document. By reviewing
these materials, you agree to not disclose these materials to any 12/053,788; US 12/055,102; US 12/057,442; US 12/057,476;
US 12/107,976; US 12/118,789; US 12/118,792; US
third party unless expressly authorized by Verint Systems, and
12/164,480; US 12/245,781; US 12/326,205; US 12/351,370;
to protect the materials as confidential and trade secret
information. Any unauthorized review, retransmission, US 12/416,906; US 12/464,694; US 12/466,673; US
12/483,075; US 12/497,793; US 12/497,799; US 12/504,492;
dissemination or other use of these materials is strictly
US 12/539,640; US 12/608,474; US 12/628,089; US
prohibited. If you are not authorized to review these materials,
please return these materials (and any copies) from where they 12/630,030; US 12/684,027; US 12/686,213; US 12/708,558;
US 12/725,127; US 12/753,137; US 12/762,402; US
were obtained. All materials found herein are provided “AS IS”
12/768,194; US 12/792,796; US 12/840,227; US 12/840,233;
and without warranty of any kind.
The Verint Systems Inc. products are protected by one or more US 12/852,144; US 12/879,868; US 12/887,059; US
12/887,089; US 12/888,445; US 12/891,620; US 12/915,868;
of the following U.S., European or International Patents: USPN
US 12/915,941; US 12/916,006; US 12/940,508; US
5,659,768; USPN 5,790,798; USPN 6,278,978; USPN
6,370,574; USPN 6,404,857; USPN 6,510,220; USPN 12/942,111; US 12/964,891; US 13/005,996; US 13/008,283;
US 13/011,870; US 13/011,871; US 13/016,998; US
6,724,887; USPN 6,751,297; USPN 6,757,361; USPN
13/036,923; US 13/096,145; US 13/096,148; US 13/096,153;
6,782,093; USPN 6,952,732; USPN 6,959,078; USPN
6,959,405; USPN 7,047,296; USPN 7,149,788; USPN US 13/114,620; US 13/149,655; US 13/155,343; US
13/182,672; US 13/187,438; US 13/189,514; and other U.S.
7,155,399; USPN 7,203,285; USPN 7,216,162; USPN
and International Patents and Patents Pending.
7,219,138; USPN 7,254,546; USPN 7,281,173; USPN
7,284,049; USPN 7,325,190; USPN 7,376,735; USPN Certain components of this Software or software contained in
this Product (collectively, “Software”) may be covered by so-
7,424,715; USPN 7,424,718; USPN 7,466,816; USPN
called “free or open source” software licenses (“Open Source
7,478,051; USPN 7,558,322; USPN 7,570,755; USPN
7,574,000; USPN 7,587,041; USPN 7,613,290; USPN Components”), which includes any software licenses approved
as open source licenses by the Open Source Initiative or any
7,633,930; USPN 7,634,422; USPN 7,650,293; USPN
similar licenses, including without limitation any license that, as
7,660,307; USPN 7,660,406; USPN 7,660,407; USPN
7,672,746; USPN 7,680,264; USPN 7,701,972; USPN a condition of distribution of the Open Source Components
licensed, requires that the distributor make the Open Source
7,734,783; USPN 7,752,043; USPN 7,752,508; USPN
Components available in source code format. A license in each
7,769,176; USPN 7,774,854; USPN 7,787,974; USPN
7,788,286; USPN 7,792,278; USPN 7,792,671; USPN Open Source Component is provided to you in accordance with
the specific license terms specified in their respective license
7,801,055; USPN 7,817,795; USPN 7,822,018; USPN
terms. EXCEPT WITH REGARD TO ANY WARRANTIES OR OTHER
7,826,608; USPN 7,836,171; USPN 7,848,524; USPN
7,853,006; USPN 7,852,994; USPN 7,853,800; USPN RIGHTS AND OBLIGATIONS EXPRESSLY PROVIDED DIRECTLY
TO YOU FROM VERINT, ALL OPEN SOURCE COMPONENTS ARE
7,853,753; USPN 7,864,946; USPN 7,873,156; USPN
PROVIDED "AS IS'' AND ANY EXPRESSED OR IMPLIED
7,881,216; USPN 7,881,471; USPN 7,882,212; USPN
7,882,217; USPN 7,885,813; USPN 7,899,178; USPN WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
7,899,180; USPN 7,899,176; USPN 7,904,481; USPN
PARTICULAR PURPOSE ARE DISCLAIMED.
7,903,568; USPN 7,904,325; USPN 7,907,142; USPN
7,913,063; USPN 7,920,482; USPN 7,925,889; USPN Any third party technology that may be appropriate or necessary
for use with the Verint Product is licensed to you only for use
7,930,314; USPN 7,949,552; USPN 7,953,621; USPN
with the Verint Product under the terms of the third party license
7,953,719; USPN 7,953,750; USPN 7,965,828; USPN
7,966,397; USPN 7,991,613; USPN 7,995,612; USPN agreement specified in the Documentation, the Software or as
provided online at http://verint.com/thirdpartylicense. You may
8,000,465; USPN 8,005,676; USPN 8,015,042; USPN
not take any action that would separate the third party
D606,983; USPN RE40,634; USPN RE41,534; USPN RE41,608;
AU 2003214926; CA 2,474,735; CA 2,563,960; CA 2,564,127; technology from the Verint Product. Unless otherwise permitted
CA 2,564,760; CA 2,567,232; CA 2,623,178; CA 2,627,060; CA under the terms of the third party license agreement, you agree
to only use the third party technology in conjunction with the
2,627,064; CA 2,628,553; EP 1096382; EP 1248449; EP
Verint Product.
1284077; DE 1284077; FR 1284077; DE 833489; FR 833489;
GB 833,489; GB 2374249; IE 84821; IE 85519; IL 13532400;
NZ 534642; ZL 200520118289.3; ZL 200520118288.9; ZL VERINT, the VERINT logo, ACTIONABLE INTELLIGENCE,
200520118287.4; USPN D606,983; USPN RE40,634; USPN POWERING ACTIONABLE INTELLIGENCE, INTELLIGENCE IN
RE41,534; USPN RE41,608; AU 2003214926; CA 2,474,735; CA ACTION, ACTIONABLE INTELLIGENCE FOR A SMARTER
2,563,960; CA 2,564,127; CA 2,564,760; CA 2,567,232; CA WORKFORCE, VERINT VERIFIED, WITNESS ACTIONABLE
2,623,178; CA 2,627,060; CA 2,627,064; CA 2,628,553; EP SOLUTIONS, VOVICI, STAR-GATE, RELIANT, VANTAGE, X-TRACT,
1096382; EP 1248449; EP 1284077; DE 1284077; FR 1284077; NEXTIVA, EDGEVR, ULTRA, AUDIOLOG, WITNESS, the WITNESS
DE 833489; FR 833489; GB 833,489; GB 2374249; IE 84821; logo, IMPACT 360, the IMPACT 360 logo, IMPROVE EVERYTHING,
IE85519; IL 13532400; NZ 534642; ZL 200520118289.3; ZL EQUALITY, CONTACTSTORE, CLICK2STAFF, BLUE PUMPKIN, GMT,
200520118288.9; ZL 200520118287.4; and other provisional and GMT Planet are trademarks or registered trademarks of Verint
rights from one or more of the following Published U.S. Patent Systems Inc. or its subsidiaries. Other trademarks mentioned are
Applications: US 10/061,491; US 10/467,899; US 10/633,357; the property of their respective owners.
US 11/166,630; US 11/345,587; US 11/359,195; US
11/359,319; US 11/359,356; US 11/359,357; US 11/359,358; Visit our website at www.verint.com/intellectualpropertynotice
US 11/359,532; US 11/361,208; US 11/388,944; US for updated information on Verint Intellectual Property.
11/394,408; US 11/394,410; US 11/394,794; US 11/395,759;
US 11/475,683; US 11/477,124; US 11/478,714; US Document Revision 1.06
11/479,056; US 11/479,267; US 11/479,506; US 11/479,899; Published November 25, 2012
US 11/509,549; US 11/509,550; US 11/540,107; US
Contents
About this Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1 Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Accessing the Speech Analytics Application . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Understanding the Speech Analytics Window. . . . . . . . . . . . . . . . . . . . . . . . . . 12
Data Analyzed by Speech Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Transcribed Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Online Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Trends Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Speech and Text Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Contact Segment Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Speech Analytics Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Start your Work Session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Contact Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
TellMeWhy™ Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Speaker Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Mixed Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Contact Playback and Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Category Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Reports and Data Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Speech Analytics Alerts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Speech Analytics Alert Triggers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Alerts Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Alert format. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Speech Analytics Trends Dashboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2 Trends and Trend Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Trends in Speech Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Trend Analysis Tab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Category Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Calculating Category Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
View Category Trends and their Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Customer Behavior Indicators™. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Calculating Term Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
How to View Customer Behavior Indicators™ and Their Analysis? . . . . . . . . . . . . . . . 27
Contents
Working in the Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Understanding the Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Displaying Numeric Values in the Plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Understanding the Dynamic Scale of the Plot . . . . . . . . . . . . . . . . . . . . . . . . 31
Adding Historical Category Trends to the Plot . . . . . . . . . . . . . . . . . . . . . . . . 33
Adding Index Data Category Trends in the Plot . . . . . . . . . . . . . . . . . . . . . . . 34
Comparing Historical and Index Data Trends . . . . . . . . . . . . . . . . . . . . . . . . 35
Adding Customer Behavior Indicators™ to the Plot . . . . . . . . . . . . . . . . . . . . . 37
Showing or Hiding Trends in the Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Printing the Plot Line and Contact Volume Charts . . . . . . . . . . . . . . . . . . . . . . 39
Understanding the Plot Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Changing the Time Frame in the Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Drilling Down from the Plot Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Removing Trends from the Plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Trend Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Introduction to Trend Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Viewing Category Trend Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Viewing Customer Behavior Indicators™ Analysis . . . . . . . . . . . . . . . . . . . . . . 44
Tips for Viewing Trend Analysis Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Trend Analysis: Closer Look . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Changing the Time Frame of Trend Analysis. . . . . . . . . . . . . . . . . . . . . . . . . 50
Searching from the Trend Analysis Tab . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Trend Analysis for Speech and Text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Generating Trend Analysis Results for Speech and Text . . . . . . . . . . . . . . . . . . . 52
Generating a Volume Source for Speech or Text . . . . . . . . . . . . . . . . . . . . . . . 53
Analyzing Text Entries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Drilling Text Search Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Adding a New Customer Behavior Indicators™ Term for Speech or Text . . . . . . . . . . . . 55
3 Customer Interaction Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

Introduction to Customer Interaction Summary . . . . . . . . . . . . . . . . . . . . . . . . 57
Customer Interaction Summary Window . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Setting a Customer Interaction Summary Date Range . . . . . . . . . . . . . . . . . . . . . 59
Analyzing the Customer Interaction Summary List . . . . . . . . . . . . . . . . . . . . . . . 60
The Speech Analytics External Player . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4 Contact Search and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Contact Analysis Tab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Home Page and Results Page of the Contact Analysis Tab . . . . . . . . . . . . . . . . . . 66
Searching in Speech Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Retrieving all Transcribed Contact Segments. . . . . . . . . . . . . . . . . . . . . . . . . . 67
Searching by Keywords or Phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Searching by Terms Spoken by Agent or Customer. . . . . . . . . . . . . . . . . . . . . . . 69
Building Text Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Unauthorized Characters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Building More Complex Text Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Guided Search. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Impact 360 Advanced Speech Analytics User Guide 4

Contents
Using Autocompletion in Text Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Context-Based Suggestions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Understanding the Term Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Filtering Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Defining Search Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Editing an Uploaded External Data File . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Deleting an Uploaded External Data File . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Setting Filters from the Results Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Filter List Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Checking if Search Filters are Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Using Categories to Find Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Searching for Contacts that Indicate Emotion and Dissatisfaction . . . . . . . . . . . . . . . . 106
Understanding Search Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Understanding the Results Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Understanding the Results List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Viewing Collapsed Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Viewing Query Results Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Working with Contact Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Searching within Found Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Refining Search Results through Different Search Methods . . . . . . . . . . . . . . . . . . . 122
Working with Saved Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Accessing Saved Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Saving a Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Renaming a Saved Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Changing a Private Saved Search to a Public Saved Search . . . . . . . . . . . . . . . . . 125
Deleting a Saved Search. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Using your Search History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
History Bar Items . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Using Charts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .127
Viewing the Number of Contact Segments in Charts . . . . . . . . . . . . . . . . . . . . . 127
Viewing Percentage Values in Charts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Starting a New Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .148
5 Contact Playback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

Playing Back Contacts in Speech Analytics. . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Understanding the Speech Analytics Player . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Speech Analytics Player without Speaker Separation . . . . . . . . . . . . . . . . . . . . . . 156
Viewing Screens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Configuring the Screen Display. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Controlling Audio Playback while Viewing Screens . . . . . . . . . . . . . . . . . . . . . . 157
Opening a Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
6 TellMeWhy Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

Viewing TellMeWhy Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
7 Category Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

Speech Analytics Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

Contents
Category Builder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

Opening the Category Builder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Understanding the Category List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Understanding the Category Definition Page . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Category Definition Page Legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Category Definition Page - Filter by Rank Level . . . . . . . . . . . . . . . . . . . . . . . 169
Creating Speech Analytics Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Defining Script Adherence and Non-Script Adherence Categories . . . . . . . . . . . . . . . . 172
Defining a Script Adherence Category . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Defining a Script Non-Adherence Category . . . . . . . . . . . . . . . . . . . . . . . . . 175
Naming Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Retrieving the Complement Set of Calls. . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Use Case Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Selecting Category Colors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .177
Defining the $ Impact of a Category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Category Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Category Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Category Term Definition: Classic vs. Graphical View . . . . . . . . . . . . . . . . . . . . 181
Switching Between Classic and Graphical View . . . . . . . . . . . . . . . . . . . . . . . 182
Defining Category Terms in Classic View . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Graphical View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .185
Enter a Term Correctly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Warning Icons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .194
Defining the Rank Threshold of a Category . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Defining Category Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Hit or No Hit Segments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
The Impact of Category Filters on Category Performance in Trends . . . . . . . . . . . . . .196
Previewing Category Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
How to Define a Category’s Filter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Defining Category States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .198
Category States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .198
Number of Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .199
The Impact of Changing a Category’s State . . . . . . . . . . . . . . . . . . . . . . . . . 200
How to Define a Category’s State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Copying Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Editing Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Synchronizing Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Deleting Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Exporting and Importing Categories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Creating Category Definition Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
8 Reports and Data Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

Reports and Data Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Creating Chart Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Creating Contact Data Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Compare Two Result Sets Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Comparing Two Result Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

Contents
Comparing Two Result Sets Report Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 217

Managing Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Saved Reports Columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Refreshing the Report Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Saving a Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .222
Deleting a Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .222
9 Ranking and Document Relevancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

Retrieving Information: Recall Vs. Relevancy . . . . . . . . . . . . . . . . . . . . . . . . . 224
Single and Multiple-Term Search Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Single Term Query Default Use Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Single Term Query Using Find Within Use Case . . . . . . . . . . . . . . . . . . . . . . . 227
Single Term Query Using IDF Use Case . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Multiple Term Query Use Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
IDF Impact in Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .228
10 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Frequently Asked Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Category Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Charts Troubleshooting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Error Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .234
Filter Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
General Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .240
Playback Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Speaker Separation Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Searches Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Trends Troubleshooting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
11 Quick Reference Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251

Building Advanced Text Queries in Speech Analytics . . . . . . . . . . . . . . . . . . . . . . 251
Building More Complex Text Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

Preface
About this Guide
The Advanced Speech Analytics User Guide describes the Advanced Speech Analytics
application and the tasks required to manage the vast repository of contact data and
generate valuable business intelligence.
This introductory section contains general information about this guide, including the
following topics:
 Intended Audience, page 8
 Documentation Feedback, page 8
 Technical Support, page 9
 Verint University, page 9
.
Intended Audience
This guide is designed to be used by:
 Business Partner professional services staff responsible for planning and setting up
systems.
 Customers’ IT staff responsible for system maintenance.
Documentation Feedback
We strive to produce the highest quality documentation products and welcome your
feedback. If you have comments or suggestions about our guides or online help, you can
email us. Please include the following information with your feedback:
Preface About This Guide
 Product name and version number

 Name of online help or guide
 Topic name and page number (if applicable)
 Brief description of content
 Your suggestion for correcting or improving the documentation
Please send your messages to userguides@verint.com.
The email address is only for documentation feedback. If you have a technical question,
please contact Technical Support.
Technical Support
Our goal at Verint Systems is to provide you with the best products backed by a high-
quality support network with a variety of resource options. Verint Systems Technical
Support services include email and telephone support.
To learn more about the Witness Actionable Solutions support options that best suit your
needs, visit us at verint.com/WASsupport.
Verint University
In addition to documentation, online Help, and support services, Verint Systems also
offers both classroom-based and online learning alternatives to suit your specific needs.
To learn more about available training options from Verint Systems, visit us at
verint.com/training.

Chapter 1
Getting Started
This chapter outlines the information required for starting to work with Speech
Analytics, by covering the following topics:
 Introduction, page 11
 Accessing the Speech Analytics Application, page 12
 Understanding the Speech Analytics Window, page 12
 Data Analyzed by Speech Analytics, page 14
 Speech and Text Analytics, page 15
 Contact Segment Transcription, page 17
 Speech Analytics Tasks, page 17
 Reports and Data Export, page 20
 Speech Analytics Alerts, page 21
 Speech Analytics Trends Dashboard, page 22
Chapter 1 - Getting Started Introduction
Introduction
The typical contact center collects and stores a vast amount of data in the form of
customer interactions. It is widely recognized that these customer interactions, contain
information about the root causes of key business issues.
Designed to deliver valuable intelligence that business users readily understand and
use, Speech Analytics can help reveal the cause/effect relationships that underlie
performance and business outcomes across the enterprise, without the complexity
usually associated with advanced analytical technologies.
By revealing both what is happening and why, Speech Analytics helps equip
organizations to make better-informed decisions, maximize strengths, address
deficiencies, and make the most of market perceptions and opportunities. For instance,
mining information from contact center calls using Speech Analytics, can be an early
warning system, before an issue escalates to negative social media.
The goal is to analyze information that can help a company improve customer service,
get reactions to new products or policies, and so on. That is, Speech Analytics can help
companies turn thousands of calls into actionable data.
The Speech Analytics application uses speech-to-text technologies to transcribe
recorded customer interactions and to transform them into a searchable database. It
combines platform-agnostic Speech Analytics with a proven methodology to help
organizations enhance customer retention and satisfaction, increase first call resolution
and improve sales and self service effectiveness.
The Speech Analytics application automatically categorizes and prioritizes transcribed
interactions based on specific business issues relevant to your contact center. The
Speech Analytics application then enables you to access the transcribed contacts for
playback, enabling you to hear the context in which the words were said and thus
identify issues critical to your business needs.
With Speech Analytics you can:
 Gain insight from recorded calls to help you improve products, processes,
competitive advantage, and the overall customer experience. See “Contact Search
and Analysis” on page 63.
 Use advanced search capabilities to research any hypothesis and quickly receive a
prioritized list of results out of millions of calls. See “Understanding Search Results”
on page 106.
 Categorize and analyze call content automatically to tell you why customers call,
reveal the root cause of customer perceptions and high call volumes, and identify
competitive threats and opportunities. See “Category Definition” on page 163.
 Surface trends that might otherwise go undetected without listening to thousands of
calls. See “Trends and Trend Analysis” on page 23.
 Share high-value intelligence across the enterprise with automated workflow tools.
See “Reports and Data Export” on page 208.

Chapter 1 - Getting Started Accessing the Speech Analytics Application
 Use enhanced Speech Analytics to analyze text. See “Trend Analysis for Speech and
Text” on page 52.
 Analyze the customer’s communication patterns with the contact center, emerging
tendencies and significant events. See “Customer Interaction Summary” on
page 56.
Accessing the Speech Analytics Application

If you have been assigned access to the Speech Analytics application and to Speech
Analytics instances, the Speech Analytics application menu appears in the Navigation
panel on the left side of the Portal home page. This menu allows you to access the
Speech Analytics application and its instances.
The Speech Analytics application may comprise one or more instances, where each
instance supports a different language or a different line of business (LOB). Instances
are created when the system is set up.
To access the Speech Analytics application:
1 From the Portal, select the Analytics option to display a menu containing the
Speech Analytics instances to which you have been assigned.
2 Under Speech Analytics, click an instance to open it.
The Speech Analytics home page opens, and the instance name appears at the top
of the page.
See “Understanding the Speech Analytics Window” on page 12.
To access a different Speech Analytics instance, you must return to the Portal home
page and select it from the Speech Analytics menu.
You cannot open more than one instance at a time from the same the Portal window. To
open a new instance, you must open another Portal window.
NOTE You can also access the Speech Analytics application by selecting Interactions >
Quality Monitoring > Speech Analytics and selecting an instance.
Understanding the Speech Analytics Window

The following is an example of the Speech Analytics application window.
See the “Legend” on page 13 for an explanation about each area.

Chapter 1 - Getting Started Understanding the Speech Analytics Window
Legend
 Toolbar
The toolbar at the top of the Speech Analytics window displays buttons that enable
you to access functionality relevant to the tasks you are performing and to the
information displayed.
When you are working in Trend Analysis tab or the Contact Analysis tab, the main
toolbar is displayed. The main toolbar enables you to access the Category Builder,
saved searches, and filtering and reporting functionalities of the application.
The buttons on the toolbar change when you view the Results page. For more
information, see “Understanding the Results Page” on page 107.
 Query Result Metrics
Click the Query Result Metrics button to display query result metrics for the
retrieved result set.
For more information, see “Viewing Query Results Metrics” on page 113.
 Search
The Search field enables you to perform searches with keywords or phrases, using
guided search tools such as, autocompletion and context-based suggestions.

Chapter 1 - Getting Started Data Analyzed by Speech Analytics
 Trends
When you open Speech Analytics, the Trend Analysis tab is displayed. For more
information, see “Trend Analysis Tab” on page 24.
You can click the Contact Analysis tab to view information about the transcribed
contact segments. For more information, see “Contact Analysis Tab” on page 64.
 Help
The Help button in the top right corner of the application window enables you to
access the application's online help.
Data Analyzed by Speech Analytics

The following sections describe the data that you can view and analyze using Speech
Analytics.
 Transcribed Contacts, page 14
 Online Data, page 14
 Trends Data, page 15
Transcribed Contacts
Out of the entire scope of the contact segments recorded for your contact center,
Speech Analytics only works with transcribed contact segments. The percentage of
contact segments to transcribe and their types are determined by transcription rules.
For more information, see “Contact Segment Transcription” on page 17.
Online Data
Transcribed contact segments and their details are stored in the Speech Analytics online
database, also referred to as the Index. All search, playback and analysis operations
(excluding trends and customer behavior indicator analysis) are performed against this
database.
The following sections describe how to work with online data, by covering the following
topics:
 Time Period Covered, page 14
 Authorized Data and Operations, page 15
Time Period Covered

The Speech Analytics online database covers data of a relatively recent time period, up
to a few months at most. This period can be extended or shortened depending on the
amount of contact segments that you transcribe. For instance, this period will be longer
if there is only one Transcription Service in the system, or shorter if there are two
Transcription Services. With regard to transcription rules, this period is shorter if you

Chapter 1 - Getting Started Speech and Text Analytics
transcribe, let's say, 100% of recorded contact segments, and longer if you transcribe
10% of your recordings.
When the maximum Speech Analytics database capacity is reached, the system
automatically starts removing older contact segments from the Speech Analytics online
database. When a contact segment is removed from the online database, it is no longer
available for searches, analysis, and playback.
Authorized Data and Operations

The data that you can view and the operations that you can perform depend on your
visibility settings and the rights assigned to you in the Assignment Manager.
For example, you can perform text searches only if you are assigned the "Search by
keywords" right in the Assignment Manager. In addition, if a search retrieves 100 results
in a keyword search, it does not necessarily mean that there are 100 segments
containing this keyword, but rather that you are authorized to view these specific
contact segments. Your access can be restricted, for example, because you have been
assigned only some of the groups and therefore you can view only their data.
This online help refers to the Assignment Manager rights and visibility settings that have
an effect on the operations you are performing or the data you are viewing.
Trends Data
To calculate trends, Speech Analytics collects statistics on category performance and
term usage and stores this data in the Speech Analytics historical database. The time
period covered by the historical database is from 18 months ago up to the most recent
date. This is the time period for which Customer Behavior Indicators™ and Category
Historical Trends are available.
NOTE Index data category trends are based on the Speech Analytics online database.
Speech and Text Analytics

Advanced Speech Analytics is integrated with Text Analytics. This integration enables
you to access the vast repository of customer interaction data, and analyze both speech
and text data in a unified manner.
You can also mine information from contact center calls using the Speech Analytics
application and from text sources (for example, Tweets, emails or Facebook entries)
using the Text Analytics project.
Text Analytics is a set of linguistic techniques that retrieve information from textual
sources for business intelligence, data analysis, research and investigation. It studies
word frequency distributions, pattern recognition, and so on, turning text into data for
analysis. Text Analytics targets and integrates feedback from internal sources (for

Chapter 1 - Getting Started Speech and Text Analytics
example, surveys, call center notes, and chat/email exchanges) and external sources
(for example, social media and other consumer generated content).
The integration of Speech Analytics and Text Analytics, enables you to:
 Perform a unifined Speech and Text Trend analysis. For more information, see
“Trend Analysis for Speech and Text” on page 52.
 Analyze different types of data for a single customer in one place. For more
information, see “Customer Interaction Summary” on page 56.
 Analyze integrated Speech Analytics results from within the Text Analytics Navigator
and Text Analytics Reports. For more information about the Text Analytics
Navigator, see the Text Analytics Reference Guide.
 Search, organise, and analyze text to understand root cause, trends and more using
the Text Analytics Navigator. For more information about the Text Analytics
Navigator, see the Text Analytics Reference Guide.
The analysis of text sources is only available with a licensed Speech Analytics version
that enables Text Analytics.

Chapter 1 - Getting Started Contact Segment Transcription
Contact Segment Transcription

The contact segments transcribed by Speech Analytics are based on rules defined to
meet your contact center's needs. These rules define the percentage of contact
segments to transcribe out of the total number of segments recorded by your system,
and may also include a description of the type of segments that should be transcribed.
For example, you may have a rule defining that 10% of the Marketing group's segments
should be transcribed. Transcription rules are created based on the Speech Analytics
application administrators.
In addition, your system may include multiple instances of the Speech Analytics
application, where each instance supports a different language or a different line of
business (LOB). Each instance of the Speech Analytics application contains its own
unique set of transcription rules and can therefore provide access to a completely
different set of contact segments. For more information about the Speech Analytics
application rules, see Storage Manager Rule Editor chapter in the Interactions and
Analytics Administration Guide.
Speech Analytics Tasks

The tasks that you can perform in Speech Analytics allow you to efficiently analyze
customer interactions with the contact center to gain insights into your company's
performance and to identify and monitor issues important for your business.
There is no strict workflow which you should follow when working in the Speech
Analytics application. Rather, your tasks should be based on your needs and on the
results of analysis performed by the application.
Speech Analytics enables you to perform the following tasks. Except for the first, all of
these tasks depend on the information obtained in the Trend Analysis and Contact
Analysis tabs:
 Start your Work Session, page 18
 Contact Search, page 18
 TellMeWhy™ Analysis, page 18
 Speaker Separation, page 18
 Contact Playback and Transcription, page 19
 Category Definition, page 20

Chapter 1 - Getting Started Speech Analytics Tasks
Start your Work Session

Your typical work session will start with reviewing information in the Trends Analysis
tab. This tab presents category performance and term frequency trends derived from
processing the information that has been accumulated so far. For more information, see
“Trend Analysis Tab” on page 24.
You may also review the information in the Contact Analysis tab, which shows the
number of contact segments in each category. This tab allows you to access saved
searches and view charts on various aspects of the contact center performance. For
more information, see “Trend Analysis Tab” on page 24.
Contact Search
You may choose to find and listen to contact segments related to a certain issue. For
instance, these can be contact segments belonging to a specific category, containing a
certain term, or handled by a specific group of people in the contact center. To enable
you to find contact segments of your interest, Speech Analytics offers several search
methods, such as guided keyword searches, search by categories, and filtering, and
allows you to combine these methods. For more information, see “Searching in Speech
Analytics” on page 66.
TellMeWhy™ Analysis
Use the TellMeWhy™ analysis to view root cause suggestions for any set of retrieved
results. For more information, see “Viewing TellMeWhy Analysis” on page 160.
Speaker Separation
Speaker Separation enables the separation of the audio of both sides of the conversation
(agent and customer). Speaker Separation occurs during the acquisition process, at the
time of contact segment recording. In Speech Analytics, this extends the search and
playback capabilities.
 New search capabilities
Speaker Separation provides additional search capabilities allowing you to search
for keywords or phrases spoken by the customer or by the agent. This ability can be
also used in category terms. For more details on this type of search, see “Searching
by Keywords or Phrases” on page 67.
When searching by keywords or phrases spoken by the customer or the agent, you
can use such guided search options as autocompletion and context-based
suggestions.
 Contact playback
When playing back contacts recorded with Speaker Separation, you can distinguish
who is talking (agent or customer) and what they are saying during their
conversation. The speaker indication is provided on the audio waveform, in speaker
bars, and in transcription as well. For more details, see “Understanding the Speech
Analytics Player” on page 151.

Chapter 1 - Getting Started Speech Analytics Tasks
For each segment recorded with Speaker Separation, the acquisition process also
provides a set of data, specific to Speaker Separation. For any single contact segment
recorded with Speaker Separation, Speech Analytics provides statistics such as the
agent or customer total talk time, percentage of agent or customer talk time, and the
number of talk-overs initiated by the agent or customer. For details, see “Viewing
Contact Details” on page 115.
With Speaker Separation statistics, you receive additional capabilities described below.
 Query Result Metrics
Speaker Separation statistics is included in query result metrics available for any
search results. Query result metrics contain Speaker Separation values similar to
those available for a single contact segment, but calculated for a given set of results
- for instance, average percentage of agent or customer talk time for a given result
set. For details, see “Viewing Query Results Metrics” on page 113.
 Filtering by Speaker Separation data
You can use Speaker Separation statistics in search and category filters. For a list of
filters, specific to Speaker Separation, see “Filter List Parameters” on page 95.
 Charts
New charts that focus on Speaker Separation statistics are provided. See “Using
Charts” on page 127 for more details.
 Reports
 Query result metrics, with Speaker Separation statistics, are provided for chart
reports.
 Contact data reports provide Speaker Separation statistics for all exported
contact segments.
Mixed Systems
With Speaker Separation enabled, your system may be working in a mixed mode, when
it contains segments recorded with Speaker Separation and segments recorded without
Speaker Separation. This can occur when the calls source is from a system where the
recording units have been configured with Speaker Separation, or when your system
was updated from earlier versions of Speech Analytics.
In such systems, Speaker Separation statistics is calculated and displayed only for
segments recorded with Speaker Separation. For example, if both types of segments
appear in the search results, the Speaker Separation data metrics are calculated only on
those segments that were recorded with Speaker Separation.
Contact Playback and Transcription

You can playback any retrieved contact segment and view its transcription to
understand the context in which certain keywords have been used, or the issues that
came up in the conversion between the customer and the agent. For more information,
see “Playing Back Contacts in Speech Analytics” on page 150.

Chapter 1 - Getting Started Reports and Data Export
Category Definition
You may need to refine category definitions in your system. See “Speech Analytics
Categories” on page 164 to learn about categories, their usage, and how to create, edit,
and delete them.
Reports and Data Export

Some of the Speech Analytics tasks are not directly related to speech analysis but they
facilitate access to this information and its sharing. These are report and export tasks
that allow converting the application data to different formats, such as Excel or PDF.
Reports can be printed, saved, and sent by email to be shared with users throughout the
enterprise. For more information, see “Reports and Data Export” on page 209.

Chapter 1 - Getting Started Speech Analytics Alerts
Speech Analytics Alerts

Speech Analytics can generate real-time alerts for specific events that occur in the
system, in the form of pop-up messages or emails that appear on the user’s
workstation.
For information on the alerts mechanism and creating alert rules, see Enterprise Suite
Framework Administration Guide.
 Speech Analytics Alert Triggers, page 21
 Alerts Flow, page 21
 Alert format, page 22
Speech Analytics Alert Triggers

Speech Analytics alerts are triggered for the following conditions:
 Speech Analytics Category Threshold
The alert is triggered when a category (active or published), exceeds a defined
threshold set for that category.
 Speech Analytics Category Hit Alert
The alert is triggered when a contact matches a specific category’s (active or
published) criteria.
 Speech Analytics Category Trend Threshold
The alert is triggered when a trend exceeds a defined threshold set for category
trend results.
 Speech Analytics Term Trend Threshold
The alert is triggered when a term exceeds a defined threshold set for the Term
Trend results.
Alerts Flow
The following steps describe how alerts are triggered and sent in Speech Analytics:
1 One or more Speech Analytics alert rules must be defined in the Tracking module,
specifying the conditions for triggering alerts in the Speech Analytics application.
2 Every five minutes, the system’s alert mechanism checks to see if a rule’s trigger
conditions are met.
3 If a condition is met, an alert is triggered in the form of an email or pop-up message
according to the rule definitions.
For each rule whose conditions are met, one alert is triggered. If more than one rules’
conditions are met, a separate alert is triggered for each rule.
The alert is triggered for the target users defined in the alert rule.
See Enterprise Suite Framework Administration Guide for information on how to define
the rules.

Chapter 1 - Getting Started Speech Analytics Trends Dashboard
Alert format
Each alert, whether it is a pop-up message or an email, contains one or more of the
following details:
 Rule Name: Appears in the email Subject, only if in the alert rule it is configured to
appear.
 Instance / Category Name
 Defined Threshold and Time Frame
 Category Usage Value: The value that triggered the alert.
 Categorized Contacts SID Number
 Categorize Contact External Player Link: A link that opens the External Player
with the contact.
 Category List: A list of categories that exceeded the threshold, including the value
that triggered the alert.
 Terms List: A list of terms that exceeded the threshold, including the value that
triggered the alert.
Speech Analytics Trends Dashboard

The suite wide web application includes performance analytics that integrate and
analyze data from a variety of systems across the enterprise. This data is presented in
dashboards to help identify areas of improvement.
Dashboards are a convenient way to display timely information in the form of graphic
reports that enable administrators to determine the causes of unacceptable performance
levels.
Speech Analytics Trends is one of the many available dashboard reports.
To create a Speech Analytics Trends dashboard and determine the data displayed by the
dashboard, you must configure or edit the following parameters through the Enterprise
Manager:
Column Description
Trends Type Select Term or Category to determine the type of trends that
will appear in the Dashboard.
Speech Instance Select the Speech Instance ID associated with the trends that
will appear in the Dashboard.
Time Period Select the time period for which you want trends to appear in
the Dashboard.
For more information about configuring and editing the Dashboard refer to the
Interactions and Analytics Administration Guide.

Chapter 2
Trends and Trend Analysis
This chapter outlines the information required for understanding customer

communication with the contact center, emerging tendencies and significant events, by
covering the following topics:
 Trends in Speech Analytics, page 24
 Trend Analysis Tab, page 24
 Category Trends, page 25
 Customer Behavior Indicators™, page 27
 Working in the Plot, page 28
 Trend Analysis, page 42
 Trend Analysis for Speech and Text, page 52
Chapter 2 - Trends and Trend Analysis Trends in Speech Analytics
Trends in Speech Analytics

Speech Analytics is a powerful tool that provides insights about the vast amount of calls
your contact center handles daily.
Trends help to:
 Identify changes in customers communication with the contact center.
 Understand any emerging tendencies.
 Pinpoint any significant events that may require close attention.
 Analyze text sources with a licensed Speech Analytics version that enables Text
Analytics. See Trend Analysis for Speech and Text, page 52.
As a result, the application can reveal critical information of which you may not be
aware. Therefore, trends as part of Speech Analytics, provide a powerful competitive
advantage, as well as the opportunity to correct process or service issues before they
escalate.
Two types of trends are calculated - Category Trend and Customer Behavior
Indicators™.
All trends are visually presented in a dedicated Plot. Alongside a graphical presentation
of each trend, Speech Analytics also provides trend analysis data. You can view this data
in time frames of various duration to receive an understanding of recent trends and
trends over a longer period of time.
All trend data is displayed in the Trend Analysis tab.
NOTE If an alert was defined to be sent when a trend or term exceeds a defined
threshold, an alert message (email or pop-up) will be sent to all defined users. For
information on alerts see “Speech Analytics Alerts” on page 21.
Trend Analysis Tab

The Trend Analysis tab is the first tab displayed when you open the Speech Analytics
application and is also referred to as the landing page. The Trend Analysis tab displays
results of Category and Customer Behavior Indicators™ analysis and enables you to
view trends in the Plot. The information you obtain from the Trend Analysis tab can often
determine your subsequent tasks in the application - for example, you may need to
investigate a steep drop in a category's performance or search by terms whose
frequency has increased lately.
NOTE If the index is empty the graph in the Trends Analysis tab will be empty and the
word “Undefined” will appear.
The following is a list of all the areas that make up the Trend Analysis tab.
 Category Trends Analysis

Chapter 2 - Trends and Trend Analysis Category Trends
The Category Trend Analysis area provides a summarized view of category trends
over the selected time frame. For more information, see “Viewing Category Trend
Analysis” on page 42.
 Customer Behavior Indicators™ Analysis
The Customer Behavior Indicators™ Analysis area displays a list of 50 terms that
showed a distinct rise or drop in their frequency during the selected time frame. For
each term, a summarized view of its trend over the selected time frame is provided.
For more information, see “Viewing Customer Behavior Indicators™ Analysis” on
page 44.
 Plot
The Plot displays line charts of chosen trends. Up to five trends can be displayed in
the Plot at one time. For more information, see “Understanding the Plot” on
page 29.
 Plot Manager
The Plot Manager displays a list of trends that have been chosen to appear in the
Plot and allows you to manage that list. For more information, see “Understanding
the Plot Manager” on page 39.
By default, the Category Trend Analysis and Customer Behavior Indicators™ Analysis
areas display trend analysis of the last seven days. The Plot displays line charts of the
three top Customer Behavior Indicators™ and shows their development in a time frame
of the last 6 weeks.
When working in the Trends Analysis tab, you can change the time frame in any of the
areas and choose to view with up to 5 trends in the Plot.
Your changes are maintained throughout your work session. The next time you open the
Speech Analytics application, the Trend Analysis tab is restored to its defaults to present
the most updated trend analysis for the last seven days, with the Plot showing the top
three Customer Behavior Indicators™ and their development during the last 6 weeks.
To learn more about the main areas of the Trend Analysis tab, refer to the following
topics:
 Viewing Category Trend Analysis, page 42
 Viewing Customer Behavior Indicators™ Analysis, page 44
 Understanding the Plot, page 29
 Understanding the Plot Manager, page 39
Category Trends
Speech Analytics categories are an efficient way of finding calls pertaining to specific
business issues among the total scope of contact segments. Once you have created
categories that correspond to business issues (for example, Customer Complaints,
Billing Issues, Product Feedback, or Repeat Calls), Speech Analytics automatically
categorizes calls according to their relevance to the issue. For more information about
Speech Analytics categories, see “Speech Analytics Categories” on page 164.

Chapter 2 - Trends and Trend Analysis Category Trends
While categorization provides quantitative information (such as the number of customer

complaints about late shipments or the number of repeat calls), category trends show
how these numbers change over time. Some categories may experience seasonal
fluctuations, others may be constant or show steady growth or decline. With trends, you
can easily learn regular patterns of your categories and immediately notice any
deviation from these patterns.
Let's assume, your system contains a category called "Billing Inquiries". Its trend shows
that the category is relatively stable but has regular monthly peaks. The rise occurs
during the first week of each month when customers receive their bills and call with
questions. After that, the category performance returns to a stable state. If there were
any deviations from this trend (abnormally high increases, increases in the middle of the
month), it would be clearly indicated in the trend, and you would be able to take
appropriate action.
This topic outlines the information required for understanding category trends, by
 Calculating Category Performance, page 26
 View Category Trends and their Analysis, page 27
From the trend, you can immediately perceive that the rise in April is much higher than
regular monthly rises (as in January, February and March). You may need to investigate
the reasons of such a steep rise. For more information about interpreting data in the
Plot, see “Understanding the Plot” on page 29.
Calculating Category Performance

In trends, a category's performance reflects the percentage of "hit" segments out of the
relevant transcribed contact segments, on a given date. In that calculation:
 Hit segments are all segments that qualify for that category as matching both the
category filter and terms.
 The relevant contacts are either all transcribed segments, or if a filter was created
for that category, all transcribed segments that match the filter. For more
information, see “Defining Category Filters” on page 195.
For example, a category's filter can limit the minimum duration of a contact segment to
3 minutes. In this case, the category's "hit" segments are all contact segments that are
3 minutes long or longer and also match the category's term definition. The category's
performance is the relation of "hit" segments to the total number of segments that are 3
minutes long or longer - both of those that do and do not match the category's terms.
Let's assume, Speech Analytics processed 1,000 contact segments on a given date.
 900 segments matched the category filter. The category performance will be based
on these contact segments.
 300 out of those 900 contact segments also matched the category's term definition.
These 300 segments are the "hit" segments of the category.
The category performance on that date is (300/900)*100% = 33.3%.

Chapter 2 - Trends and Trend Analysis Customer Behavior Indicators™
View Category Trends and their Analysis

Speech Analytics allows you to view trends of any published and active category that
you select. For each category, two types of trends are distinguished: historical and Index
data. Trends are shown as line charts in the Plot. To view any of these trends, you have
to add it to the Plot. For more information, see “Adding Historical Category Trends to the
Plot” on page 33 and “Adding Index Data Category Trends in the Plot” on page 34.
In addition, category trend analysis is provided. This information is displayed in the
“Viewing Category Trend Analysis” on page 42 area of the Trend Analysis tab.
Customer Behavior Indicators™

As Speech Analytics processes transcribed contact segments, it automatically identifies
terms that come up in calls and collects their statistics. A term can be:
 A single word - for example, price
 A two- or three-word phrase - for example, "price increase", "recent price increase"
Based on the collected statistics, Speech Analytics can calculate trends of any of the
terms. A term's trend shows the frequency with which that term came up in
conversations, on any given date.
This topic outlines the information required for understanding Customer Behavior
Indicators™, by covering the following topics:
 Calculating Term Frequency, page 27
 How to View Customer Behavior Indicators™ and Their Analysis?, page 27
Calculating Term Frequency

Term frequency is the percentage of contact segments in which the term is used, out of
the total number of contact segments on a given date.
For example, on a given date, a term was used in 250 out of 1,000 contact segments.
The term frequency for that date is (250/1,000)*100%=25%.
How to View Customer Behavior Indicators™ and Their

Analysis?
When you open Speech Analytics, the Trend Analysis tab (the landing page of the
application) displays trends of three terms that have the highest frequency change in
the last seven days. The trends are shown as line charts in the Plot, in a 6 week time
frame. The following is an example of the initial view of the Plot in the landing page.

Chapter 2 - Trends and Trend Analysis Working in the Plot
In the example, terms with the highest frequency change in the last seven days are
"disconnect", "disconnected" and "server". For more information about interpreting data
in the Plot, see “Understanding the Plot” on page 29.
You can view trends of any other terms as well. To view Customer Behavior Indicators™,
you need to add it to the Plot. For more information, refer to “Adding Customer Behavior
Indicators™ to the Plot” on page 37.
In addition, Customer Behavior Indicators™ analysis is provided. This information is
displayed in the Customer Behavior Indicators™ Analysis area of the Trend Analysis tab.
For more information, see “Viewing Customer Behavior Indicators™ Analysis” on
page 44.
Working in the Plot

This topic outlines the information required for understanding the Speech Analytics
Plots, by covering the following topics:
 Understanding the Plot, page 29
 Displaying Numeric Values in the Plot, page 30
 Understanding the Dynamic Scale of the Plot, page 31
 Adding Historical Category Trends to the Plot, page 33
 Adding Index Data Category Trends in the Plot, page 34
 Comparing Historical and Index Data Trends, page 35
 Adding Customer Behavior Indicators™ to the Plot, page 37
 Showing or Hiding Trends in the Plot, page 38
 Printing the Plot Line and Contact Volume Charts, page 39
 Understanding the Plot Manager, page 39
 Changing the Time Frame in the Plot, page 40
 Drilling Down from the Plot Manager, page 40
 Removing Trends from the Plot, page 41

Understanding the Plot

The Plot allows you to view any Category or Customer Behavior Indicators™ as a line
chart. You can see the trend's development over the selected time frame and easily
identify periods when the trend was consistent and periods when it was falling or rising.
By changing the time frame of the Plot, you can zoom in and out on periods of interest.
Up to five trends can be displayed in the Plot at a time.
To help you evaluate the trend, the Plot also displays the number of processed contact
segments for each day.
NOTE Adobe Flash Player version 9 or above is needed for viewing the Plot. If you do not
have the required version of the player, the Plot is not displayed. Instead, a link
from which you can download the player is displayed.
This topic outlines the information required for understanding the Plot, by covering the
following topics:
 Time Frame, page 29
 Line Charts, page 29
 Processed Calls, page 29
 Time Axis, page 30
Time Frame
The time frame of the Plot is shown in the top right corner of the Plot.
Line Charts
In the upper area, you can view trends' line charts.
 Each trend appears in a different color, as indicated in the Plot Manager. For more
information, see “Understanding the Plot Manager” on page 39.
 The scale of this area of the Plot is determined by the highest value of the displayed
trends, over the selected time frame. For more information, see “Understanding the
Plot Manager” on page 39.
Processed Calls
The lower area displays the number of processed contact segments for the selected time
frame. For each single date, a separate contact volume bar is displayed. The bar shows
the number of transcribed contact segments on that day. No bars are displayed for dates
on which no segments were transcribed.
The scale of this area depends on the highest contact volume displayed, while the width
of the bars is determined by the time frame of the Plot. For more information, see
“Understanding the Dynamic Scale of the Plot” on page 31.

Time Axis
The upper and the lower areas of the Plot are synchronized with the time axis displayed
at the bottom of the Plot. For any date in the Plot, you can see both the trend value and
the number of contact segments.
Below the time axis, you can see fields and buttons for defining the time frame of the
Plot. For more information, see “Changing the Time Frame in the Plot” on page 40.
Depending on the time frame, vertical grid lines in the Plot can indicate different points
in time.
Time Frame Vertical grid lines mark...
Longer than a month The first day of each month.
One month The first day of each week of the month.
One week and shorter A day in the week.
Displaying Numeric Values in the Plot

When a trend is displayed in the Plot, you can see its line chart without numeric values.
To display numeric values in the Plot:

1 Hover your mouse over the Plot. The pointer changes to the hand symbol, and the
following values are displayed:
 In the top right corner of the Plot, the date range is replaced with the date to
which you are pointing.
 Percentage of contact segments for each displayed trend appears in the top left
corner of the Plot. The color of each value corresponds to its trend's color in the
Plot.
TIP Check the colors of the displayed trends in the Plot Manager. For more information,
see “Understanding the Plot Manager” on page 39.
 In the lower area of the chart, the number of contact segments processed on
that date is shown; the contact volume bar of that day is marked with a dot.
2 To view values of a different date, move the mouse horizontally along the Plot. If
you move the mouse from the Plot, the values are no longer displayed.
NOTE If no contact segments were processed on a certain date, line charts continue
without gaps but trends' values for that date are not available.

Understanding the Dynamic Scale of the Plot

Both the upper and the lower areas of the Plot have dynamic scales that are
automatically adjusted to the highest value displayed.
This topic outlines the information required for understanding the Plot, by covering the
following topics:
 Line Charts, page 31
 Contact Volume, page 32
Line Charts
For example, if the highest value of the trend is 4%, then the highest available value on
the scale in the upper area of the Plot will be 4% as well.
The line chart scale can change when you:

 Remove the trend with the highest value from the Plot. For more information, see
“Removing Trends from the Plot” on page 41.
 Add a trend with a higher maximal value. For more information, see “Adding
Historical Category Trends to the Plot” on page 33, “Adding Index Data Category
Trends in the Plot” on page 34, “Adding Customer Behavior Indicators™ to the Plot”
on page 37.
 Show or hide the trend with the highest value. For more information, see “Showing
or Hiding Trends in the Plot” on page 38.
 Change the time frame of the Plot. As a result, a different highest value for the
trend is displayed. For more information, see “Changing the Time Frame in the Plot”
on page 40.
When the scale changes, the appearance of all trends in the Plot changes as well. For
example, if you add a trend with a higher maximum value, the yellow trend, shown in
the example above, will be displayed with smaller amplitudes.

Contact Volume
The scale in the lower area of the Plot changes in accordance with the highest daily call
volume for the time period displayed.
In the following example, you can see that the maximal value in the scale changed from
20K to 40K when a different time period was selected in the Plot.
Additionally, the width of the volume bars in the lower area of the Plot depends on the
defined time frame. The shorter the time frame, the wider the bars. In the image below,
compare the bar widths for a period of a week and for a period of 3 months.

For more information about defining time frames in the Plot, see “Changing the Time
Frame in the Plot” on page 40.
Adding Historical Category Trends to the Plot

Historical category trends are based on the Speech Analytics historical database. They
show the category's performance for each given date during the last 18 months up to
the most recent date.
The Speech Analytics historical database stores statistics on category performance and
term usage. This database is used for calculating Customer Behavior Indicators™ and
Category Historical Trends. The time period covered by the historical database is from
18 months ago up to the most recent date.
New data is added to the historical database daily, but no updates of previously inserted
data occur. For that reason, the historical trend always shows the category's
performance as it was calculated on that date.
 When a new category is created, its historical trend starts from the date of its
creation and continues up to the most recent date. For more information, see
“Creating Speech Analytics Categories” on page 170.
 Modifying a category's definition does not affect its historical trend prior to the
modification, but only from the modification point onward. Therefore, a historical
trend of a category is expected to show a change, sometimes an abrupt one, on
that date. For more information, see “Editing Categories” on page 202.
 When a category is deleted, its historical trend is deleted. For more information, see
“Deleting Categories” on page 204.
 If an active or published category was deactivated, and then activated again later,
there will be a gap in the historical trend of the category for the period of time when
the category was inactive. For more information, see “Defining Category States” on
page 198.
NOTE Unlike historical trends, Index data trends are entirely updated upon any change in
the category definitions. For more information, see “Adding Index Data Category
Trends in the Plot” on page 34 and “Comparing Historical and Index Data Trends”
on page 35.
To add a historical category trend to the Plot:

1 In the Trend Analysis tab, in the Category Trend Analysis table, locate the
required category.
2 If the category does not appear on the first page of the list, use the pagination
buttons at the bottom of the list to go through the list. For more information, see
“Tips for Viewing Trend Analysis Data” on page 46.
3 Click the category name. The trend is added to the Plot Manager. In the Plot
Manager, the trend is marked with the icon. The color of the category's line chart
is automatically selected and is not consistent with the color assigned to the
category in the Category Builder. The trend's line chart, in the selected color, is also
displayed in the Plot.

NOTE If the trend is added to the Plot Manager but is not shown in the Plot, see
“Showing or Hiding Trends in the Plot” on page 38.
Adding Index Data Category Trends in the Plot

Categories' Index data trends are based on the Speech Analytics online database and
last for the entire period of time covered by this database (up to a few months). These
trends are dynamically updated with any changes that you make in category definitions.
The Speech Analytics online database, also referred to as the Index, stores transcribed
contact segments and their details. All search, playback and analysis operations in
Speech Analytics are performed against the Speech Analytics online database. If a
contact segment is no longer available in the online database, it cannot be retrieved in
searches or played back. Speech Analytics online database stores data of a relatively
recent time period, up to a few months at most.
The ability of Index data trends to dynamically reflect any occurring changes makes
them a helpful tool in creating and editing categories. With Index data trends, you can
immediately see what the performance of a new category would be, or how changing a
category's definition affects its performance.
 When you create a new category, all contact segments that are in the online
database and that qualify for this category are retroactively assigned to it. The
Index data trend of the new category can be immediately viewed in the Plot as if the
category existed in the system previously. For more information, see “Creating
Speech Analytics Categories” on page 170.
 When you edit a category's terms or filter, all contact segments available in the
online database are scanned, and the relevant ones are assigned to the category.
Contact segments that are no longer relevant for the category are disassociated
from it. Upon a change in the category's definition, its Index data trend is entirely
updated and can be completely different from what it was before the change. For
more information, see “Editing Categories” on page 202.
 When you delete a category, its Index data trend is no longer available. For more
information, see “Deleting Categories” on page 204.
NOTE In the historical trend, any change in category definitions will be reflected from the
date of the change only. For more information, see “Adding Historical Category
Trends to the Plot” on page 33 and “Comparing Historical and Index Data Trends”
on page 35.
To add an Index data category trend to the Plot:

1 In the Trend Analysis tab, in the Category Trend Analysis table, locate the
required category.
2 If the category does not appear on the first page of the list, use the pagination
buttons at the bottom of the list to go through the list. For more information, see
“Tips for Viewing Trend Analysis Data” on page 46.
3 Click the Options button to the left of the category name.

4 Select Add to Plot > [Category name] (Index data).

For example, to add the Index data trend of the Customer Experience category,
select Add to Plot > Customer Experience (Index data). The trend is added to
the Plot Manager. In the Plot Manager, the trend is marked with an icon. The color
for the category's line chart is automatically selected and is not consistent with the
color assigned to the category in the Category Builder. The trend's line chart, in the
selected color, is also displayed in the Plot.
Comparing Historical and Index Data Trends

The differences between historical and Index data trends are explained by the databases
on which these trends are based:
 Historical trends are based on the Speech Analytics historical database that covers
up to 18 months.
 Index data trends are based on the Speech Analytics online database that can cover
up to a few months and is dynamically updated.
For a period of time that is covered by both databases, both historical and Index data
trends are available. These trends are almost identical if the category definition remains
the same during that period of time, or can be different if the category definition was
changed.
In the following examples, historical and Index data trends of the same categories are
shown. The historical database starts on Jan 01, 2008, and the online database starts
on Feb 05, 2008 in all examples.
Example 1
The following example shows the historical and Index data trends of the Customer
Education category. The green line represents the Index data trend, the red line shows
the Historical trend.
The historical trend starts from Feb 05, 2008, from the same date when the historical
database starts. The Index data trend is available starting from Jan 01, 2008, according
to the earliest date in the online database.
The Customer Education category definition did not change over the time period covered
by the online database. For that reason, the category's historical and Index data trends
almost coincide.
The parts of the line charts that fully coincide appear in one color only. The color is
automatically chosen out of the two colors used for the category's trends. In parts which
are different, you can see both colors.

Example 2
In the following image, you can see the historical and Index data trends for the
Competition category, which was changed on Feb 20, 2008. The blue line shows the
Index data trend; the red line shows the Historical trend.

A change in the category's definition caused the following changes in its trends:
 In the historical trend, a steep rise occurred on the date of the change - from 0%
to 9% on average. After that date, the category's performance remained relatively
constant, with periodic drops and rises.
 The entire Index data trend was updated, because the category was retroactively
applied to all contact segments in the online database. The Index data trend
demonstrates the same pattern as the historical trend after the modification date -
with the average value of 9% and periodic drops and rises.
The historical trend of the category starts from the date it was created and first contact
segments were assigned to it; its Index data trend starts from the earliest date in the
online database, because the category was retroactively applied to all contact segments
in the online database. Note that the historical trend requires some period (typically, a
day) to stabilize. During that period, the historical trend is approaching the Index data
trend. Once the historical trend reaches the Index data trend, both trends show the
same pattern.
Adding Customer Behavior Indicators™ to the Plot

You can add the trend of any term to the Plot. These can be terms that appear in the
Trend Analysis tab, for example, in the Term Analysis area, or most impacting terms of
category trends.
If the term whose trend you want to view does not appear in the Trend Analysis tab, you
can type it in and then view its trend.

This topic outlines the information required for understanding how to add customer
behavior indicators to the plot, by covering the following topics:
 Adding a Customer Behavior Indicator™ to the Plot, page 38
 Adding New Customer Behavior Indicators™, page 38
Adding a Customer Behavior Indicator™ to the Plot

1 In the Trend Analysis tab, locate the required term. The term's trend is added to
the Plot Manager and displayed in the Plot.
2 The order of terms in the list is determined by their relative change. If the term
does not appear on the first page of the list, use the paginations buttons at the
bottom of the list to go through the list. For more information, see “Tips for Viewing
Trend Analysis Data” on page 46.
3 Click the required term. The trend is added to the Plot Manager. In the Plot Manager,
the trend is marked with the corresponding icon , and a color for its line chart is
automatically selected. The trend's line chart, in that color, is also displayed in the
Plot.
Adding New Customer Behavior Indicators™

1 In the Trend Analysis tab, in the Plot Manager area, enter the new term in the
Add to Plot field.
As you start typing the term, the autocompletion list displays the possible options
for completing the word. The list is based on the terms available in the Speech
Analytics historical database. These are the terms whose trends you can view.
2 Select the relevant word from the list. If you type in a term that does not appear on
the list, the trend will not be displayed.
3 Click the Add to Plot button to the right of the field. The trend is added to the
Plot Manager. In the Plot Manager, the trend is marked with the corresponding icon
, and a color for its line chart is automatically selected. The trend's line chart, in
that color, is also displayed in the Plot.
Showing or Hiding Trends in the Plot

The Plot can display any trend that was added to the Plot Manager but not more than
five trends at a time. If the Plot is already displaying five trends, you need to first hide
at least one of them before additional trends can be displayed.

You can also hide one or more trends in the Plot to have a clearer view of a certain
trend. Trends that were hidden can be displayed in the Plot again at any time.
1 In the Trend Analysis tab, in the Plot Manager, review the list of trends. A
selected check box indicates that the trend is displayed in the Plot.
2 Clear the check boxes of the trends that you want to hide in the Plot.
3 Select the check boxes of the trends that you want to show in the Plot.
Printing the Plot Line and Contact Volume Charts

To print a hard copy of the Plot Line and Contact Volume charts:
1 Right-click the Line Chart or Contact Volume area of the Plot.
A small menu appears.
2 Select Print.
3 Select a Printer and click Print.
Understanding the Plot Manager

The Plot Manager appears to the right of the Plot in the Trend Analysis tab. It displays a
list of trends that have been chosen to appear in the Plot and enables you to manage
that list.
From the Plot Manager, you can obtain the following information on each trend:
 The selected check box to the left of the trend name indicates that the trend is
shown in the Plot. If the check box is cleared, the trend is hidden.
 The icon to the right of the trend's name indicates the type of trend, which can be:
 - Historical Category Trend
 - Index Data Category Trend
 - Customer Behavior Indicators™
 The colored square ( ) indicates the trends’ color in the Plot
From the Plot Manager, you can:
 Adding Customer Behavior Indicators™ to the Plot, page 37
 Showing or Hiding Trends in the Plot, page 38
 Removing Trends from the Plot, page 41
 Drilling Down from the Plot Manager, page 40

Changing the Time Frame in the Plot

There are multiple ways of changing the time frame in the Plot. You can:
 Using Default Time Ranges, page 40
 Specify the Exact Date Range in the Plot, page 40
 Marking the Time Frame on the Time Axis, page 40
NOTE Changing the time frame of the Plot can cause changes in its scale. For more
information, see “Understanding the Dynamic Scale of the Plot” on page 31.
Using Default Time Ranges

In the Trend Analysis tab, in the lower right corner of Plot, click the Zoom button of
the time frame you want to set for the Plot.
Specify the Exact Date Range in the Plot

In the Trend Analysis tab, in the lower left corner of Plot, enter the exact dates in the
Custom Period fields. The dates must be entered in the MM-DD-YYYY format.
Marking the Time Frame on the Time Axis

1 Point with the mouse to the right or left end of the Plot's time axis. Little handles
appear at both ends of the time axis.
2 Drag one of the handles to the required date on the time axis. As you drag the
handle, the time range of the Plot narrows according to the dates between the
handles, and the line charts are updated accordingly. The exact date range is shown
in the top right corner of the Plot.
3 If needed, drag the other handle to a different location on the time axis to adjust
the time frame to the required duration.
4 To use the defined time frame as a sliding time frame, place the mouse between the
two handles and drag the time frame along the time axis.
As you drag your time frame, the appearance of trends in the Plot changes, and the
date range in the top right corner of the Plot is updated accordingly.
Drilling Down from the Plot Manager

You can drill down on any Customer Behavior Indicators™ displayed in the Plot Manager
to retrieve contact segments that are relevant for the trend.

The drill-down is similar to performing a search with the following search criteria:
 If you are drilling down on a Customer Behavior Indicators™, the term is regarded
as a search keyword.
 If you are drilling down on a Category Trend, then only contact segments belonging
to the category are retrieved. If you are drilling down on an Index data category
trend, these are all the segments on which this trend is based.
Additionally, the time frame of the Plot is used as a date range search filter. For instance,
if the time frame of the Plot is one month, then the retrieved segments will belong to the
last 30 days only.
NOTE Remember that contact segments are retrieved from the online database, which
covers up to a few months. If the time frame of the Plot is wider (for example, a
year), drill down retrieves contact segments of a specific date range - where the
time frame of the Plot overlaps the date range of the online database. Drill-down
retrieves no results if the time frame of the Plot does not overlap the date range of
the online database (for example, if the Plot displays data of more than half a year
ago).
The Speech Analytics online database, also referred to as the Index, stores
transcribed contact segments and their details. All search, playback and analysis
operations in Speech Analytics are performed against the Speech Analytics online
database. If a contact segment is no longer available in the online database, it
cannot be retrieved in searches or played back. Speech Analytics online database
stores data of a relatively recent time period, up to a few months at most.
Drill Down to Contact Segments from the Plot Manager

In the Plot Manager, click the Drill Down to Contacts icon to the left of the trend.
The contact segments are displayed in the Results page. For more information, see
“Understanding the Results Page” on page 107.
You can also perform keyword searches for terms appearing in the Trend Analysis tab.
For more information, see “Searching from the Trend Analysis Tab” on page 51.
Removing Trends from the Plot

You can remove any trend from the Plot to free space for trends which you wish to add.
In the Trend Analysis tab, in the Plot Manger area, click the Remove from Plot
button next to the trend you want to remove. The trend is removed from the Plot
Manager. If the trend was displayed in the Plot, it is removed from the Plot as well.
TIP You can also hide trends in the Plot without removing them from the Plot Manager.
For more information, see “Showing or Hiding Trends in the Plot” on page 38.

Chapter 2 - Trends and Trend Analysis Trend Analysis
Trend Analysis
This topic outlines the information required for understanding the Trend Analysis, by
 Introduction to Trend Analysis, page 42
 Viewing Category Trend Analysis, page 42
 Viewing Customer Behavior Indicators™ Analysis, page 44
 Tips for Viewing Trend Analysis Data, page 46
 Trend Analysis: Closer Look, page 47
 Changing the Time Frame of Trend Analysis, page 50
 Searching from the Trend Analysis Tab, page 51
Introduction to Trend Analysis

In addition to presenting trends as line charts, Speech Analytics performs complex
analysis of each trend and enables you to view the results of this analysis over different
time frames.
All trend analysis data is displayed in the Trend Analysis tab. For more information, see
“Viewing Category Trend Analysis” on page 42 and “Viewing Customer Behavior
Indicators™ Analysis” on page 44.
NOTE Trend analysis returns results only if the number of contact segments for the
selected time period is sufficient. For minimal data thresholds of trend analysis,
see “Changing the Time Frame of Trend Analysis” on page 50.
For each trend, three numeric parameters - average value, absolute change, and
relative change – are calculated for any selected time frame. These values reflect a
trend's development during that time frame and highlight any changes that may have
occurred. To learn more about these parameters, see “Trend Analysis: Closer Look” on
page 47.
When viewing trend analysis data, you can:
 Display any trend in the Plot to view it in greater detail. For more information, see
“Adding Historical Category Trends to the Plot” on page 33, “Adding Index Data
Category Trends in the Plot” on page 34, and “Adding Customer Behavior
 Change the time frame of trend analysis. For more information, see “Changing the
Time Frame of Trend Analysis” on page 50.
Viewing Category Trend Analysis

Category trend analysis provides a summarized view of category trends over the
selected time frame. Results of the analysis are displayed in the Category Trend Analysis
area of the Trend Analysis tab.

NOTE Category trend analysis is based on historical trends only. For more information,
see “Adding Historical Category Trends to the Plot” on page 33.
At the top of the list, you can see categories with the highest absolute change in
performance, over the selected time frame.
By default, the Category Trend Analysis area displays analysis results for a Weekly trend
type based on the current date. You can select a different trend type (for example, Daily,
Monthly, etc.) and a different date by clicking the Settings icon ( ) in the top right
corner of the Category Trend Analysis area. For more information, see “Changing the
Time Frame of Trend Analysis” on page 50.
The Category Trend Analysis header contains details about the time period for which the
trends are displayed. This information is updated every time the Settings are changed.
A blue asterisk (*) in the Category Trend Analysis header indicates that there is no data
for the selected date. The data is retrieved for the closest date to the selected date.
The following columns are displayed in the area:
Column Description
Options button Use the Options button to do the following:

 Add the category's historical and Index data trends to the Plot.
For more information, see “Adding Historical Category Trends to
the Plot” on page 33.
 Add trends of the most impacting terms to the Plot. For more
information, see “Adding Index Data Category Trends in the
Plot” on page 34.
 Add the most impacting terms to the Search field. For
information, see “Searching from the Trend Analysis Tab” on
page 51.
Category The category name.

Click the category name to add the category’s historical trend to
the Plot.
For more information, see “Adding Historical Category Trends to
the Plot” on page 33.
Up and Down arrows The arrow indicates the trend direction over the selected time
frame and is displayed only for daily and weekly trend analysis.
Average Shows the average category performance (in percentage) over the
selected time frame. In daily analysis, the value in the Average
column is the actual performance of a category on that day.

Column Description
Change The Change column shows the absolute and relative changes in the
category's trend in the following format: absolute change (relative
change %). The methods used to determine the absolute change
and the relative change are different in short-term and long-term
analysis, therefore the values in the Change column depend on the
selected time frame. For a detailed explanation of the Change
column, see “Trend Analysis: Closer Look” on page 47.
The absolute change determines the sorting order of categories in
the list, whereby categories with the highest absolute change in
performance appear at the top of the list. The direction of the
change does not affect the sorting order. For example, a category
with the absolute change in performance of -40% will be placed
above a category with the absolute change of +30%.
Impacting Terms Displays two terms of the category that had the highest impact on
the category's trend over the selected time frame. These are two
terms with the highest relative change over the selected time
frame.
Click a term to add its trend to the Plot. For more information, see
“Adding Customer Behavior Indicators™ to the Plot” on page 37.
See “Trend Analysis: Closer Look” on page 47 to better understand the meaning of the
Average and Change columns.
In addition, see “Tips for Viewing Trend Analysis Data” on page 46.
Viewing Customer Behavior Indicators™ Analysis

Customer Behavior Indicators™ analysis provides a summarized view of Customer
Behavior Indicators™ over the selected time frame and determines 50 terms with the
highest change in frequency over that time frame. Results of the analysis are displayed
in the Customer Behavior Indicators™ Analysis area of the Trend Analysis tab.
NOTE The selection of terms is based not on the frequency itself but on the change in it.
Any most commonly used term will not appear among results of Customer
Behavior Indicators™ analysis, if its frequency remains relatively constant over the
selected time frame. Only if a term's frequency has changed considerably, the
term can be included into the 50 terms brought by the Customer Behavior
Indicators™ analysis.
At the top of the list, you can see terms with the highest absolute change in frequency,
over the selected time frame.
By default, the Customer Behavior Indicators™ Analysis area displays analysis results
for a Weekly trend type based on the current date. You can select a different trend type
(for example, Daily, Monthly, etc.) and a different date by clicking the Settings icon
( ) in the top right corner of the Customer Behavior Indicators™ Analysis area. For
more information, see “Changing the Time Frame of Trend Analysis” on page 50.

The Customer Behavior Indicators™ Analysis header contains details about the time
period for which the trends are displayed. This information is updated every time the
Settings are changed.
A blue asterisk (*) in the Customer Behavior Indicators™ Analysis header indicates that
there is no data for the selected date. The data is retrieved for the closest date to the
selected date.
The following columns are displayed in the area:
Column Description
Options button Use the Options button to do the following:

 Add trends of the term and its correlated terms to the Plot. For
more information, see “Adding Customer Behavior Indicators™
to the Plot” on page 37.
 Add the term and its correlated terms to the Search field. For
information, see “Searching from the Trend Analysis Tab” on
page 51.
Term Click the term to add its trend to the Plot. For more information,
see “Adding Customer Behavior Indicators™ to the Plot” on
page 37.
Up and Down arrows The arrow indicates the Customer Behavior Indicators™ direction
over the selected time frame and is displayed only for daily and
weekly trend analysis.
Average Shows the average term frequency (in percentage) over the
selected time frame. In daily analysis, the value in the Average
column is the actual frequency of a term on that day.
Change The Change column shows the absolute and relative changes in the
term's trend in the following format: absolute change (relative
change %). The methods used to determine the absolute change
and the relative change are different in short-term and long-term
analysis, therefore the values in the Change column depend on the
selected time frame. For a detailed explanation of the Change
column, see “Trend Analysis: Closer Look” on page 47.
The absolute change determines the selection of terms and the
sorting order of the list, whereby terms with the highest absolute
change appear at the top of the list. The direction of the change
does not affect the sorting order. For example, a term with the
absolute change of -40% will be placed above a term with the
absolute change of +30%.
Correlated Terms Displays the two top correlated terms, over the selected time
frame.
Click a term to add its trend to the Plot. For more information, see
“Adding Customer Behavior Indicators™ to the Plot” on page 37.
See “Trend Analysis: Closer Look” on page 47 to better understand the meaning of the
Average and Change columns.
In addition, see “Tips for Viewing Trend Analysis Data” on page 46.

Tips for Viewing Trend Analysis Data

 By default, trend analysis results are displayed for a week's time period, including
today's date and 6 days prior to today's date. Weekly trends show how the last
week is different from the couple of weeks that precede it.
 You can change the trend analysis time period by selecting one of the following
values in the Time Period field for a specific date:
- Day
- Week
- Month
- 3 months
- 6 months
- 9 months
- 12 months
For example, when you select October 12th in the Date field and Month in the Time
Period field, the resulting trend results are relevant for a month before and including
October 12th.
The Time Period field is displayed by clicking the Settings icon ( ) in the top right
corner of the Trend Analysis areas.
For more details, refer to “Changing the Time Frame of Trend Analysis” on page 50.
 Category Trend Analysis trends are updated once an hour.
 Customer Behavior Indicator trends are updated once a day. This update occurs as
part of the index build.
 The Customer Behavior Indicators list includes the top 50 trends arranged in pages.
Use the pagination buttons in the lower right corner of the area to move from page
to page.
 The Change column shows two values:
- Absolute change
- Relative change
The format of the Change column is absolute change (relative change %), for
example, +2.22 (31.7%) or -1.06 (-6%).
 The list is sorted by the absolute change. The direction of the change does not affect
the sorting order. For example, a trend with the absolute change of -40% will placed
above a trend with the absolute change of +30%.
 Each column has a short and a full name. Short column names are displayed as
column heading in the trend analysis area. You can view the full name of a column
in a tooltip when hovering the mouse of the column's heading.
 Click the question mark to view a detailed explanation of this area.

Trend Analysis: Closer Look

Trend analysis provides a summarized view of trends over the selected time frame. By
default, trend analysis results are calculated based on the current date and a week
before the current date. Results are updated every day, so for example, trend results
using these default settings on tomorrow's date may be different than today's trend
results based on the same default settings.
Alternatively, you can also view trend results based on a specific date in history and
within a different time period (Week, Day, Month, etc.). This allows you to view trend
results for that specific time in the past, without having it be affected by daily trend
updates.
So, for example, you can view trend results based on today’s date and a week
beforehand, and trend results based on September 3rd and a month beforehand.
With regard to the time frame duration, two types of trend analysis can be
distinguished: short-term and long-term.
 Short-term analysis is performed for daily (last day) and weekly (from a selected
date) time frames. The analysis shows the deviation of the trend's current average
from its previous average. For example, the weekly Customer Behavior Indicators™
analysis shows how the average frequency of the term in the last seven days is
different from its average frequency in a previous time period.
 Long-term analysis is performed over time frames from a month up to a year from
the current date or a selected date. The analysis shows the trend's deviation from a
baseline, while the baseline is the average value of the trend across the entire
period.
Both short-tem and long-term trend results are relevant for the selected Time Period
before and including the selected date.
For more details, refer to “Changing the Time Frame of Trend Analysis” on page 50.
For each trend, three numeric parameters - average value, absolute change, and
relative change – are calculated for any selected time frame. These values reflect a
trend's development during that time frame and highlight any changes that may have
occurred.
A trend's average value is shown in the Average column, and the absolute and relative
changes are shown in the Change column. To understand trend analysis results, let's
take a close look at the Average and Change columns and see how these values are
derived.
This topic outlines the information required for understanding the summarized view of
trends over a selected time frame, by covering the following topics:
 Average Column, page 48
 Change Column, page 48

Average Column
The Average column shows the average trend value over the selected time frame. In
daily analysis, the value displayed in the Average column is the actual trend value on
that day.
Change Column
The Change column shows the absolute and relative changes in the trend in the
following format: absolute change (relative change %). The methods used to
determine the absolute change and the relative change are different in short-term and
long-term analysis, therefore the values in the Change column depend on the selected
time frame.
The absolute change determines the sorting order of trends in the list, whereby trends
with the highest absolute change appear at the top of the list. The direction of the
change does not affect the sorting order. For example, a trend with the absolute change
-40% will be placed above a trend with the absolute change of +30%.
Change Column in Short-Term Analysis (Daily and Weekly)

 Absolute change - the difference between the trend's current average value and
its average value in the predefined preceding time period.
The absolute change can be positive or negative, according to the direction of the
trend over that period of time. If a trend's average value over the current period is
lower than over the preceding time period, the absolute change is negative. The
direction of the trend is indicated by the up and down arrows ( ).
 Relative change - the relation of a trend's average value in the previous time
period compared to the trend's current average value.
Assuming a trend's average value over the last week is 20% while its average value over
the previous period was 25%, then:
 Absolute change = current average value – previous average value (20% - 25% =
-5%)
 Relative change = absolute change/previous average value (-5%/25% =-20%)
In this case, the value in the Change column is -5 (-20%).
Change Column in Long-Term Analysis (from Month to Year)

 Absolute change - the difference between the maximal and minimal value of a
trend for the selected time period.
In long-Customer Behavior Indicators™ Analysis, the absolute change is always
positive, and no arrows indicating the direction of the change are displayed.
 Relative change - the relation of the absolute change to the average value of the
trend.

Assuming a trend's average value over the last 6 months is 20% while its highest
performance during this period was 26% and the lowest was 15%, then:
 Absolute change = highest value – lowest value (26% - 15% = 11%)
 Relative change = absolute change/average (11%/20%)=55%)
In this case, the value in the Change column is 11 (55%).
Long-term analysis detects any considerable rises or drops that occurred in a trend. To
view the exact events in a trend, add it to the Plot. For more information, see “Adding
Historical Category Trends to the Plot” on page 33 and “Adding Customer Behavior
When displayed in the Plot, trends with the same absolute change can show different
patterns, for example:
 A steady drop or rise
 A one-time rise
 Seasonal peaks
Plot Trends Example

The following image shows three trends with a similar absolute change (approximately
45%) in a 3-month time frame. In spite of their similar absolute change, the trends
reveal different patterns.
For more information on interpreting data in the Plot, see “Understanding the Plot” on
page 29.

Changing the Time Frame of Trend Analysis

By default, the Trend Analysis tab displays Category and Customer Behavior Indicators™
Analysis of the last seven days from the current day. This data is displayed in the
Category Trend Analysis and Customer Behavior Indicators™ Analysis areas of the tab,
respectively. In any of these areas, you can select any other predefined historical time
frame (for example, a specific date, weekly, monthly or yearly).
With regard to the time frame selected, two types of trend analysis can be
distinguished: short-term (daily and weekly) and long-term (from a month up to a
year). The method used to determine the value of the Change column is different with
each approach.
 Short-term Customer Behavior Indicators™ analysis compares the recent behavior
of a trend with its behavior during the preceding time period. For example, a term's
trend analysis for a week can show how the frequency of a certain term changed
during the last 7 days, in comparison with the previous time period.
 Long-term Customer Behavior Indicators™ Analysis, focuses on changes that
occurred within that period of time. For example, a term's trend analysis for a year
shows the difference between the lowest and the highest frequencies of that term in
the course of the last year.
For more information about calculations, see “Trend Analysis: Closer Look” on page 47.
NOTE Trend analysis returns results only if the number of contact segments for the
selected time frame is sufficient and only during time periods in which trend
history is available.
The minimal data thresholds are:
 Daily analysis - 3 weeks' worth of data.
 Weekly analysis – 4 weeks' worth of data.
 Long-term analysis - from a month to a year’s worth of data of the selected
time period. For example, for monthly trend analysis, a month's worth of data
is required.
When selecting data for trend analysis, Speech Analytics excludes days with
exceptionally low call volumes. Days with less than 700 calls are discarded, as well
as days in which the number of calls is significantly lower than the median number
of calls from the date when the tracking started.
1 Click the Trend Analysis tab, at the top of the Category Trend Analysis or
Customer Behavior Indicators™ Analysis area.
2 Click the Settings icon ( ) in the top right corner.
The screen is replaced with a Time Period configuration screen.
3 In the Time Period field, select the time range on which to base trend results (Day,
Week, Month, etc.).
For example, to view trend results for a three-month time period, select 3 Months.
NOTE You cannot select a date and time period for which Trend history is not available.

4 In the Date field, select the required value.

For example, when you select October 12th in the Date field and Month in the Time
Period field, the resulting trend results are relevant for a month before and including
October 12th.
5 Click Apply.
The list reloads, with the following changes:
 The list of terms is updated and can be completely different from the previously
displayed terms list.
 The correlated terms of the same term can be different.
In Category Trends Analysis:
 The order of categories can change for each selected time frame.
 Most impacting terms of each category can be different.
For more information on trend analysis, see “Viewing Category Trend Analysis” on
page 42 or “Viewing Customer Behavior Indicators™ Analysis” on page 44.
Searching from the Trend Analysis Tab

From the Trend Analysis tab, you have quick and convenient access to the search
functionality of Speech Analytics. You can search by any term that appears in the Term
Analysis area - both by terms whose trends are displayed in the list and by their
correlated terms. In addition, you can also search by the most impacting terms of
category trends.
To search for contact segments from the Trend Analysis tab:

1 In the Trend Analysis tab, in the Category Trend Analysis or Customer
Behavior Indicators™ Analysis area, click the Options button , select Add to
search, and then click the required term. The term is displayed in the Search field
above the tab.
2 Repeat the previous step for any other terms that you want to include in your
search query.
3 Click Find. The list of contact segments that contain the terms is displayed in the
Results page. For more information on the Results page, see “Understanding the
Results Page” on page 107.
NOTE Searching from the Trend Analysis tab is different from drilling down from the Plot
Manager. The search finds all the contact segments containing the term, while
drilling down finds only contact segments that belong to the time frame of the Plot
and that also meet additional search criteria. For more information, see “Drilling
Down from the Plot Manager” on page 40.

Chapter 2 - Trends and Trend Analysis Trend Analysis for Speech and Text
Trend Analysis for Speech and Text

This section includes procedures for performing a combined trend analysis of Speech
and Text:
 Generating Trend Analysis Results for Speech and Text, page 52
 Generating a Volume Source for Speech or Text, page 53
 Analyzing Text Entries, page 54
 Drilling Text Search Results, page 54
 Adding a New Customer Behavior Indicators™ Term for Speech or Text, page 55
The system enables you to select the data source for analysis of speech, text or unified
text and speech.
For a view of the Speech Analytics window without Text Analytics see “Understanding
the Speech Analytics Window” on page 12.
Generating Trend Analysis Results for Speech and Text

Both the Category Trend Analysis and Customer Behavior Indicators™ results list
contain data directly related to the selected data source.
The Customer Behavior Indicators ™ results list contains data directly related to the
selected data source.
The Data Source drop-down list is found in each of the Trend Analysis tab areas (that is,
Category Trend Table, Add to Plot, and Customer Behavior Indicators, Volume
Graph).
Trend analysis provides a summarized view of trends over the selected time frame and
data source. You can change the data source by clicking the Settings icon in the top
right corner of the Category Trend Analysis or Customer Behavior Indicators area. When
you click the Settings icon in one of these areas, the current screen is replaced with a
Source, Period and Date configuration screen. If you select a Text source, the Time
Period and Date fields are disabled. For more information, see Changing the Time Frame
of Trend Analysis, page 50.
The Source drop-down list changes according to the type and number of text source(s)
defined for your project. That is, numerous text options can be displayed for the
different text source options (for example, facebook, twitter, email, and so on).
A single Text source option along with the Speech and All options, indicates that the
data from all of the different Text Analytics sources have been grouped into one text
type.
Generating Trend results for a Text Source can create a Trend results list that contains
Linguistic Connections. A Linguistic Connection is an established connection between
two words. For example, the connection HOTEL -> GOOD can be specified for the
following sentences:
Overall a very good hotel, good location for shopping, very helpful staff, very clean.
My 2-nd best hotel recommendation.

Selecting any of the options in the Source drop-down list, will result in a list of trends
for the selected source and time period.
To Generate Trend Analysis results for Speech and Text:

1 In the Trend Analysis tab, select the area for which you would like to generate
trend results (that is, Category Trend Analysis or Customer Behavior Indicators™).
2 Click the Settings icon ( ) in the selected area.
3 From the Source drop-down list, select one of the following data sources types:
 Speech
Selecting Speech as the source of data, generates Trend Analysis results that are
only related to call center data.
 Text Source
Selecting a Text source as the source of data, generates Trend Analysis results
that are only related to the selected text source (for example, surveys, call
center notes, and chat/email exchanges, social media and other consumer
generated content).
NOTE If you select a Text source, the Date field is disabled.
 All
Selecting All as the source of data, generates Trend Analysis results that contain
both Speech and Text data.
For example, when you select All as the source and the word eat appears for
email, chat and call, the Trend Analysis results list will have three entries for the
word eat (eat<email>, eat<chat> and eat<Speech>). The brackets indicate the
data source of each trend entry.
NOTE When you select a specific source type (text or speech), the brackets do not
appear in the Trend Result list.
Analysis results are displayed in the Category Trend Analysis or Customer Behavior
Indicators™area of the Trend Analysis tab.
For additional information about the main areas of the Trend Analysis tab, see “Trend
Analysis” on page 42.
Generating a Volume Source for Speech or Text

The lower area of the graph in the Trend Analysis tab, shows the amount of processed
data for each day. With the addition of Text Analytics you can now select to show the
number of calls processed or the number of analyzed text entries for the selected
timeframe.

A Volume Source drop-down list has been added to enable you to view call and/or
document volume per source. You can change the Volume Source selection at anytime
and the Volume bar is redrawn automatically without affecting the graph above.
To generate a volume source for Speech or Text:

1 From the Volume Source drop-down list in the lower area of the Trend Analysis
Plot, select one of the following volume source types:
 Speech Source
Selecting Speech as the volume source, shows the number of processed contact
segments for the time frame.
 Text Source
Selecting a Text source as the volume source, shows the number of processed
text entries for the time frame (for example, surveys, call center notes, and
chat/email exchanges, social media and other consumer generated content).
Analyzing Text Entries

The Options button in the Category Trends Analysis or Customer Behavior Indicators™
area of the Trends tab has an additional function. In addition to adding terms to the
Search field and the Add to Plot list, the Options button enables you the ability to select
the text term or category from the Trends results and subsequently access the Text
Analytics Navigator for further text analysis.
To Analyze Text Entries:

1 In the Trend Analysis tab, select Category Trend Analysis or Customer Behavior
Indicators™.
2 Click the Options button next to the category or text term you want to analyze.
3 Click Analyze Text and the text term or correlated term you would like to analyze.
4 The Text Analytics Navigator is activated for the selected category or text term.
For more information about the Text Analytics Navigator, see the Text Analytics
Reference Guide.
Drilling Text Search Results

With Speech Analytics you can drill down to obtain detailed information about a specific
category or text entry in the Add to Plot list.
To drill down to a specific category or text entry:

1 In Trend Analysis > Add to Plot click the Drill Down to Contacts icon next to
the Category or Text entry for which you want to receive in depth information.
The Text Analytics Navigator opens.

If you click the Drill Down to Contacts icon next to a Speech Analytics entry, the
relevant contact segments are displayed in the Results page. For more information, see
Drilling Down from the Plot Manager, page 40.
Adding a New Customer Behavior Indicators™ Term for

Speech or Text
With Text Analytics, selecting a Source type in the Add to Plot area, determines the
type of term options that are displayed in the autocompletion list.
1 From the Source drop-down list in the Add to Plot list, select a source type (text or
speech).
2 Enter the new term in the Add to Plot field.
As you start typing the term, the autocompletion list displays the possible options
for completing the word. The list is based on the terms available in the Speech or
Text Analytics database. These are the terms whose trends you can view.
3 Select the relevant word from the list. If you type in a term that does not appear on
the list, the trend will not be displayed.
4 Click the Add to Plot button to the right of the field. The trend is added to the Plot
Manager. In the Plot Manager, the trend is marked with the corresponding icon, and
a color for its line chart is automatically selected. The trend's line chart, in that
color, is also displayed in the Plot.

Chapter 3
Customer Interaction
Summary
This chapter outlines the information required for understanding an individual

customer’s experience, by covering the following topics:
 Introduction to Customer Interaction Summary, page 57
 Customer Interaction Summary Window, page 57
 Setting a Customer Interaction Summary Date Range, page 59
 Analyzing the Customer Interaction Summary List, page 60
Chapter 3 - Customer Interaction Summary Introduction to Customer Interaction Summary
Introduction to Customer Interaction

Summary
Customers use a variety of channels to communicate with and talk about your business.
The Customer Interaction Summary window provides a unified perspective (per
customer), of the customer’s communication tendencies. For example, a customer may
call, email and discuss your business. With this information, you can detect and leverage
the customer’s opinion for decision making.
Customer Interaction Summary retrieves and displays customer's interactions records
from one or more of the following data sources:
 Calls from a Speech Analytics instance.
 Text channels of communication.
 Calls from the QM database.
 Additional Providers
The Customer Interaction Summary is only available with a licensed Speech Analytics
version that enables Text Analytics.
Customer Interaction Summary Window

From the Results page in the Contact Analysis tab, you can select to drill-down to a
unified view of a single customer’s interactions.
If an interaction in the Results list is associated with a Customer ID, the specific
interaction will have a Customer Interaction Summary button ( ) that when
clicked, launches the Customer Interaction Summary window for the specific customer.
From the Customer Interaction Summary window:
 Setting a Customer Interaction Summary Date Range, page 59
 Analyzing the Customer Interaction Summary List, page 60
The following is an example of the Customer Interaction Summary window:

Chapter 3 - Customer Interaction Summary Customer Interaction Summary Window
The initial Customer Interaction Summary window display, highlights the row for the
contact that was used to launch the window and displays a range of approximately 14
days before and after the date of the specific contact.
The contact used to launch the Customer Interaction Summary window is displayed and
initially highlighted in the Customer Interaction Summary.
The Customer Interaction Summary window contains the following details:
 Interactions for Customer
The Customer ID associated with the interaction.
Time Period
With the Time Period buttons, you can filter the contact list according to a specific
date range.
 Play button
The Play button launches the associated interaction player or text navigator.
 Date
The Date column displays the interaction’s timestamp, including date and time.
The Customer Interaction Summary list is sorted from the oldest to the newest
contact according to the interaction’s timestamp.
 Source
The Source column displays the type of data source associated with the specific
contact center interaction.
 A Speech source will display the words Speech Calls.
 A Quality Monitoring source will display the words Recorded Calls.

Chapter 3 - Customer Interaction Summary Setting a Customer Interaction Summary Date Range
 A Text source will display the text source (for example, email).
 Categories
The Categories column lists the categories associated with the customer’s
interaction.
 Agent
For Speech, Quality Monitoring and Text sources, the Customer Interaction
Summary window displays the name of the agent associated with the interaction.
A text source interaction, will only include the name of the agent if it has been
configured to do so.
 Summary
The summary in the Customer Interaction Summary window includes the following
information about the currnet interactions list in the window:
 Total number of interactions for customer
The total number of interactions retrieved for the specific customer ID across all
source types, for the current Customer Interaction Summary window filter.
 Earliest interaction date
The date of the most recent interaction retrieved from amongst the various
interaction types, for the current Customer Interaction Summary window filter.
 Latest interaction date
The date of the oldest interaction retrieved from amongst the various interaction
types, for the current Customer Interaction Summary window filter.
Setting a Customer Interaction Summary

Date Range
The Customer Interaction Summary window provides a Time Period control that allows
you to set a time range for the filtered interactions.
To modify the Customer Interaction Summary Time Period:

1 In the Customer Interaction Summary window, click the Show by range: radio
button.
2 Cick the Calendar button on the left and select the beginning of the time period.
3 Click the Calendar button on the right and select the end of the time period.
4 Click the Refresh button.
A list of the customer’s interactions for the selected time period appears.
NOTE To display all of the customer’s interactions (that is, to disable the selected time
period filter), click Show all and Refresh.

Chapter 3 - Customer Interaction Summary Analyzing the Customer Interaction Summary List
Analyzing the Customer Interaction

Summary List
In the Customer Interaction Summary window you can perform an in depth analysis of
each specific interaction, by viewing the interaction (for example, email, twitter, contact,
etc) in its source system.
To analyze a specific Customer Interaction Summary entry, click the Play button to
the left of the interaction. Depending on the entry selected, one of the following players
or navigator is launched:
 Speech Analytics External Player
When you click the Play button for a Speech Analytics contact segment entry, the
Speech Analytics External Player is launched.
For additional information, see the “The Speech Analytics External Player” on
page 60 and the Speech Analytics External Player API.
 Source-Highlighter
When you click the Play button for a text interaction, the Text Analytics Source-
Highlighter is launched.
For additional information, see the Text Analytics Reference Guide.
 Quality Monitoring Player
When you click the Play button for a Quality Monitoring interaction, the Quality
Monitoring Player is launched.
For additional information, see the Quality Monitoring Online Help.
 Third-Party Provider
When you click the Play button for a third-party provider’s interaction, the third-
party provider’s player is launched.
For additional information, see the relevant provider’s documentation.
The Speech Analytics External Player

From within the Customer Interaction Summary window, you can click a Play button to
launch the Speech Analytics External Player to replay a Speech Analytics contact.
The External Player is launched in an Internet Explorer browser or a pop-up dialog
window. Session metadata, associated transcript and category list information for the
specific call are displayed within the External Player window.
The following is an example of the Speech Analytics Exteral Player Window:

Chapter 3 - Customer Interaction Summary The Speech Analytics External Player
The Speech Analytics External Player window contains the following details:
 Call Summary
The Call Summary area of the External Player contains the following information
about the contact being played:
 Keywords found
Displays the category keywords associated with the specific contact.
 Emotion
Indicates whether or not the contact was categorized as emotional. For example,
for contacts that indicated a dissastifaction.
 Categories
 Displays a list of all of the categories associated with the specific contact. The
selected category is bold.
 Agent
Displays the name of the agent associated with the specific contact.
 Local Start Time
Displays the local start time for the specific contact.
 Screen Indicator
Indicates whether or not the contact is associated with a screen.
 Duration
Indicates the duration of the contact.

Chapter 3 - Customer Interaction Summary The Speech Analytics External Player
 Copy URL
Copies the URL to the External Player for the specific contact to the clipboard.
 Replay Call - Transcript
Displays the contact’s spoken words and highlights them as they are spoken when
the contact is played.
The player contains the same functionalities as the standard Speech Analytics
player.
 Call Categories
The Call Categories area of the External Player contains a list of the categories
associated with the contact being played.
The Call Categories area of the External Player is interactive. Selecting a category in
the Call Categories area, causes the information in the Call Summary area and the
term bubbles and term highlights in the Transcription area to change accorgingly.
The information shown in these two areas represents the connection between the
selected category and the specific contact.
 Call Metadata
Displays a list of the following metadata associated with the speicific contact:
 General Contact Details
Displays standard data associated with the specific contact. For example,
Number of Holds, Number of Transfers, Direction, ANI, DNIS, etc.
 Custom Data
Displays data associated with the specific contact that was configured specifically
for the customer.
NOTE If Speaker Separation is enabled and the system contains Speaker Separation
metadata, then Speaker Separation metadata will appear in the Call Metadata
area.

Chapter 4
Contact Search and Analysis
This chapter outlines the information required for understanding how to search and
analyze contact information, by covering the following topics:
 Contact Analysis Tab, page 64
 Searching in Speech Analytics, page 66
 Retrieving all Transcribed Contact Segments, page 67
 Searching by Keywords or Phrases, page 67
 Searching by Terms Spoken by Agent or Customer, page 69
 Building Text Queries, page 70
 Guided Search, page 77
 Filtering, page 91
 Using Categories to Find Contacts, page 104
 Searching for Contacts that Indicate Emotion and Dissatisfaction, page 106
 Understanding Search Results, page 106
 Searching within Found Results, page 121
 Refining Search Results through Different Search Methods, page 122
 Working with Saved Searches, page 122
 Using your Search History, page 125
 Using Charts, page 127
 Starting a New Search, page 148
Chapter 4 - Contact Search and Analysis Contact Analysis Tab
Contact Analysis Tab

The following is a sample Contact Analysis tab.
For information about each area in this tab see “Legend” on page 64.
Legend
 Top Bar
The top bar of the Contact Analysis tab provides a summary of the information
displayed on the page.
For example, when no search filters have been set you can see how many
transcribed contact segments are available.
If an filters have been set, this area indicates so and allows you to preview the
filters. For more information, see “Checking if Search Filters are Set” on page 104.
 Categories
The Categories area displays a list of categories that you can use for contact search
and analysis. These are published and active categories only.
NOTE You can view a complete list of categories, including inactive categories as well, in
the Category Builder. For more information, see “Understanding the Category List”
on page 166 and “Defining Category States” on page 198.

Chapter 4 - Contact Search and Analysis Contact Analysis Tab
The number of contact segments belonging to each category appears by the

category name. The number is dependent on the filters that have been set. With no
filters, the number represents the total number of contacts in the category. For
example, Customer Complaints (35) entry means that there are 35 contacts
segments categorized as customer complaints in the system. Yet, if you set a filter,
for example, to view contact segments of the last 7 days only, this number may go
down as it will represent contacts categorized as customer complaints in the last 7
days.
NOTE The numbers by category names reflect those specific segments that you are
authorized to view according to your rights and visibility setting in the Assignment
Manager. For more information, see “Data Analyzed by Speech Analytics” on
page 14
In this area, categories are organized in alphabetic order. If you want to see
categories sorted by the number of contact segments in each of them, view the
Category chart. This chart is displayed by default in the bottom right corner of the
tab.
The default Categories area of the Contact Analysis tab includes the following:
 Emotions:Customizable category that contains contacts the are characterized
by their emotional content.
The Emotions category is defined when your system is set up. Like other
categories, the Emotions category can be comprised of terms, filters and a dollar
value.
For more information on this area, refer to “Speech Analytics Categories” on
page 164 and “Using Categories to Find Contacts” on page 104.
 My Searches
The My Searches area displays names of the searches that you saved and allows
you to access these searches. For more information about saved searches, see
“Working with Saved Searches” on page 122.
 Graphs
The bottom of the screen displays charts of all transcribed contact segments
retrieved in the Results page.
By default, this area displays the Date chart and the Category chart. See “Using
Charts” on page 127 to learn about charts and ways of working with them.
If a chart does not appear, see “Troubleshooting” on page 229.
If no search filters have been set, the Contact Analysis tab displays information on all
transcribed contact segments that you are authorized to view according to your rights
and visibility settings in the Assignment Manager. If any filters have been set, the
content of the tab is filtered accordingly, and the displayed information relates only to
those segments that meet the filters. For more information, see “Checking if Search
Filters are Set” on page 104 and “Using Charts” on page 127.

Chapter 4 - Contact Search and Analysis Searching in Speech Analytics
Home Page and Results Page of the Contact Analysis Tab

The Contact Analysis tab can accommodate one of the following pages - Speech
Analytics home page and Results page.
The image in the “Contact Analysis Tab” on page 64 with the Categories and Saved
Searches areas illustrates the home page. This is the page from which most of your
operations in Speech Analytics start.
After you performed a search or applied filters, the appearance of the Contact Analysis
tab changes and the home page is replaced with of the Results page. For more
information, see “Understanding the Results Page” on page 107.
TIP When the home page is not displayed, the Home link appears in the top right
corner of the window, enabling you to return to this page at any time.

Searching in Speech Analytics

The Speech Analytics application enables you to use several search methods to find
contact segments of your interest:
 Guided Search - Searching by Keywords or Phrases, page 67 and use Using
Autocompletion in Text Queries, page 77 and Context-Based Suggestions, page 80
to assist you in selecting the most efficient search terms.
 Defining Search Filters, page 93
 Retrieving all Transcribed Contact Segments, page 67
The search results are displayed in the Results page in two formats:
 List - detailed contact segment list. For more information, see “Understanding the
Results List” on page 109 and “Understanding the Results Page” on page 107.
 Charts - present the contact segment information graphically, in bars. Each bar in
the chart focuses on some specific contact characteristics, for example, the date, or
duration. The charts can serve as another way of narrowing your search results as
they allow you to drill down on the bars to different subsets of contacts within the
results. For more information, see “Using Charts” on page 127.
You can gradually refine your search by “Searching within Found Results” on page 121
and “Refining Search Results through Different Search Methods” on page 122.

Chapter 4 - Contact Search and Analysis Retrieving all Transcribed Contact Segments
To facilitate your search process, the Speech Analytics also offers a few convenient ways
of accessing your previous search results. These include:
 Saving your searches to run them multiple times without the need to redefine the
search criteria. For more information, see “Working with Saved Searches” on
page 122.
 Using the search history to return to previously accessed results. For more
information, see “Using your Search History” on page 125.
You can also start a new search, related to a different issue, at any time. For more
information, see “Starting a New Search” on page 148.
Retrieving all Transcribed Contact Segments

You can retrieve all transcribed contact segments without defining any search criteria.
 Click the Find button in the Speech Analytics home page, without entering
keywords or setting filters. This retrieves a list of all available transcribed contact
segments.
The Results list will only display contacts associated with groups (and their subgroups)
that were assigned to you in the Assignment Manager.
The Results list will only display contacts associated with Contact Data and Customizable
Contact Data values that were assigned to you in the Assignment Manager.
The Results page lists contact segments with no category affiliation at the top of the list.
The Show percentage of total button is not available for charts in the Results page
in such cases, as the results represent 100% of the available contacts.
For more information about results, see “Understanding the Results Page” on page 107.
Searching by Keywords or Phrases

You can search for contact segments by keywords or phrases that were used in the
conversation between the agent and the customer. The keywords and phrases you are
searching for are often referred to as terms.
Searching by keywords or phrases involves submitting a text query in the Search field.
Then, the application presents the segments that include the text query and ranks them.
The text query that you submit can be of various complexity - ranging from a single
word to sophisticated expressions that define not only the words to search for, but also
their retrieval conditions.
For example, if you are interested in retrieving all contact segments where customers
complain about access problems to a new online service you recently launched, you can
search for the word “login”, and the Speech Analytics application will retrieve all
segments that contain that word. You can also search for a phrase, for example, “cannot
log in”, to retrieve contacts related to the same problem.

Chapter 4 - Contact Search and Analysis Searching by Keywords or Phrases
Building a more complex query requires the use of operators that allow you to set
conditions of the search. For a detailed description of all text search options, see
“Building Text Queries” on page 70.
NOTE The mechanism for calculating search results varies, depending on whether the
search entry is a single-term query, or a multiple-term query:
 For single-term queries, the system provides a list of all contacts that make the
rank threshold in terms of their overall score, regardless of the search term's
frequency.
 For multiple-term queries, the system bases its search logic on the industry-
standard Inverse Document Frequency (IDF) factor when calculating search
results. In order to maximize the relevancy of multiple-term search entries,
IDF diminishes the weight of terms that occur very frequently, and increases
the weight of terms that occur rarely.
For more information on IDF and the way Speech Analytics calculates search
results, see Ranking and Document Relevancy, page 223.
You can greatly improve the efficiency of your query by using autocompletion and
context-based term suggestions.
TIP When you playback a retrieved contact segment, all occurrences of your keywords
or phrases are underlined in the transcription area of the Speech Analytics player.
These underlined items serve as analysis and navigation aids. They immediately
draw your attention to the occurrences of your keywords and phrases in the
conversation. You can also move to relevant places in the conversation by clicking
these underlined items. For more information, see “Playing Back Contacts in
Speech Analytics” on page 150.
You can only perform text searches if you have been assigned the Search by keywords
right in the Assignment Manager.
Searches by keywords or phrases retrieve prioritized results, where each retrieved
contact segment is ranked according to the keywords and phrases it contains. The
higher the rank the more relevant that contact segment is for the search. Contact
segments are sorted by rank, while the rank can range from 2 to 5 stars. Contact
segments with one-star rank are excluded from results.
To search for contact segments by keywords or phrases:

1 In the Search field, type the text to search for. See “Building Text Queries” on
page 70 for a detailed explanation on various options of building your text query in
the Search field.
2 As you enter words to search for, an automatic word completion list appears under
the Search field. Click any relevant word to include it in your text query.
In addition to serving as a spelling aid, Term Autocompletion helps you identify
similar terms to the one you entered and provides valuable information on their
usage. To learn more, see “Using Autocompletion in Text Queries” on page 77.
3 To receive context-based suggestions for the term in the Search field, click
Visualize or More suggestions. To learn more about context-based suggestions,
see “Context-Based Suggestions” on page 80.

Chapter 4 - Contact Search and Analysis Searching by Terms Spoken by Agent or Customer
4 Click the Find button,

When searching from the Results page, click one of the following buttons:
 Click New Find to start a new search.
 Click Find Within Results to continue searching within the results that were
found in the previous search.
The Results page appears displaying a list of contact segments that contain the text you
searched for and have at least a two-star rank. If any filters were set before you
submitted your search, the results are filtered accordingly. For additional information,
see “Checking if Search Filters are Set” on page 104.
For further details about search results, see “Understanding the Results Page” on
page 107.
If you wish to further refine your search, you can continue “Searching within Found
Results” on page 121 while combining the term search with “Refining Search Results
through Different Search Methods” on page 122.
You can use the Speech Analytics application's “TellMeWhy Analysis” on page 159 to
view more suggested search terms to use.
Searching by Terms Spoken by Agent or

Customer
Speaker separation enables you to search for segments by a specific term spoken by
either the agent or the customer. Use the A: operator to build text queries that search
for segments where the term was spoken by the agent, or the C: operator if you want to
search for segments in which the term was spoken by the customer.
The ability to search for segments by terms spoken by the agent or the customer comes
in addition to the regular search by terms that retrieves all segments with the term,
regardless of the speaker.
1 In the Search field of the home page or Results page, type the required operator: A:
to search by terms spoken by the agent, or C: to search by terms spoken by the
customer.
2 Type the text for which you want to search after the colon. There should be no
space between the colon (:) and the next character.
3 Click the Find button when searching from the home page, or click the New Find
or Search within results button when searching from the Results page. The
Results page appears, displaying a list of segments that contain text you searched
for.
Example:

Chapter 4 - Contact Search and Analysis Building Text Queries
When searching from the Speech Analytics home page:

 To search for segments in which the agent said the word “help”, in the Search field,
type Help and click Find. The Results page appears, displaying a list of segments in
which the agent said “help”.
 To search for segments in which the customer said the word “help”, in the Search
field, type C:help and click Find. A list of segments in which the customer said
“help” appears.
 To search for occurrences of the word “help” regardless of the speaker, perform a
regular text search without any operators - in the Search field, type “help” and click
Find. A list of segments in which the word “help” was detected appears.
To better understand results of searches with Speaker Separation operators, notice the
following:
 Searches for a specific term said by the agent or customer also retrieve segments in
which that term was spoken during the talk-over time when the agent and the
customer were talking simultaneously.
 For segments recorded without Speaker Separation, the search retrieves all the
segments that contain the term you searched for.
You can combine the A: and C: operators with all other operators available for text
queries. For a detailed description of all text search options, see “Building Text Queries”
on page 70.
You can use Speaker Separation operators in category term definitions as well.
Segments that qualify for such a category are those where the term is spoken by the
respective participant of the conversation.
Example:
If a category's term is C:”thank you”, the category will include only those segment
where the phrase “thank you” was spoken by the customer. Segments where the phrase
“thank you” was detected in the agent's part only will not belong to this category. For
more information on defining category terms, see “Defining Category Terms in Classic
View” on page 182.
For more information on using Speaker Separation in Speech Analytics, see “Speaker
Separation” on page 18.
Building Text Queries

Text queries are used for term searches and for category terms definitions. Building
complex text queries requires the use of operators, which allow you to target specific
concepts and to specify retrieval conditions. Using the Auto Completion and Context-
Based Suggestions features improves query efficiency and can be combined with the
use of operators. Your text query can be a single word or a phrase.
Searching with operators refines your query and allows enhanced search capabilities
such as:

 Combining words and phrases

 Searching for those terms in the call opening or closing.
 Excluding contacts with a certain word.
 Applying more importance to contacts with a certain word.
 If Speaker Separation has been enabled, searching for a certain term spoken by the
agent or by the customer.
For more information, see “Searching by Terms Spoken by Agent or Customer” on
page 69.
The following are rules and examples of building text queries. For additional examples of
compound text queries, with several operators and word combinations, see “Building
More Complex Text Queries” on page 76.
Task: Definition Example: Search results

Syntax: include:
Retrieve contacts Enter the word of account Contacts where the

containing a word of interest. word “account” was
interest used.
Retrieve contacts Enter your phrase “close my account” All contacts where
containing a certain enclosed in double “check book” the phrase “close my
phrase or compound quotes. account” was used.
word (consisting of When searching for a
two words) compound word (a
word made up of two
words, like
“checkbook”), you
need to determine
whether the Speech
Analytics application
transcribes it as one
or two words. When
you start typing the
first few letters of the
word, the
autocompletion list
will indicate whether
“checkbook” has
been identified as one
word. It is always
recommended to also
try searching for the
word as a single word
and check the results.
If you receive no
results, try searching
for both words as a
phrase (like “check
book”).


Syntax: include:
Retrieve contacts Enter words and angry annoyed Any contact that
containing any of the phrases of your contains either the
words or phrases of interest. word “angry” or the
interest There is no need to word “annoyed”, or
separate the words both.
with a comma. This type of text
queries is not valid in
the definition of
category terms. If
you want to add both
words as terms to a
category, add them
one by one.
Alternatively, you can
build an expression
with an operator to
combine these words
into one term, for
example, angry AND
annoyed.
Retrieve contacts Enter the words angry AND annoyed Any contact that
containing all of the connected with the contains both the
words of interest AND operator. word “angry” and the
word “annoyed”.
Retrieve contacts Use the NOTIN credit NOTIN “credit Contacts containing
containing a keyword operator before the card” the word “credit” but
but excluding its context which is not when used as
usage in specific irrelevant for your part of the phrase
contexts, which are search. “credit card”. Note
not relevant that the search may
still return contacts
containing:
 The word
“card” when
not used as
part of the
phrase “credit
card”.
 The phrase
“credit card”, in
case the
contacts also
contain the
word “credit”
not as part of
that phrase.


Syntax: include:
Exclude contacts Add the minus sign credit -card Contacts containing
containing a specific (-) before the word. the word “credit”,
keyword. excluding the
contacts where the
word “card” was
identified.
If you are using the
minus operator, your
query must also
include at least one
other keyword or
phrase to search for
(without the minus
sign). Queries that
include only terms
with the minus sign
are not valid and
cannot be processed
(for example, - card).
Do not use the minus
sign (-) in category
terms. In a category
definition, this
operator is equivalent
to adding a term to
Avoid Terms.
Retrieve contacts Enter the words close NEAR account Contacts containing
where two words connected with the phrases such as
appear close to each NEAR operator. “close my account”,
other (from 0 to 3 “close your account”,
words apart). “close the account”,
“close my recently
opened account” etc.
Note that the order
you enter the
keywords defines the
phrases that will be
searched. In the
example entering
“close NEAR account”
will not return
contacts with
“account” appearing
before “close” such as
“take into account
that if you close it”.


Syntax: include:
Attribute more Add the plus (+) bill +credit Contacts where the
importance to a sign before the word. word “bill” or “credit”
certain word and thus is found, but contacts
have contacts with with the word “credit”
that word appear at may be ranked higher
the top of your search in the search result
results list. list.
Do not use the plus
sign (+) in category
terms. In a category
definition, this
operator is equivalent
to adding a term to
Very Important
Terms.
Retrieve contacts in Type A: and then the A:welcome Contacts where the
which certain words word, phrase or word “welcome” was
of phrases were expression you are spoken by the agent
spoken by the agent looking for. There (during the agent talk
should be no space time) and also during
between the colon (:) the talk-over time
and next character. when the agent and
the customer were
speaking at the same
time.
See “Searching by
Terms Spoken by
Agent or Customer”
on page 69 for more
details on this type of
search.
Retrieve contacts in Type C: and then the C:"thank you “Contacts where the
which certain words word, phrase or phrase “thank you”
of phrases were expression you are was spoken by the
spoken by the looking for. There customer (during the
customer should be no space customer talk time)
between the colon (:) and also during the
and next character. talk-over time when
the agent and the
customer were
speaking at the same
time.
See “Searching by
Terms Spoken by
Agent or Customer”
on page 69 for more
details on this type of
search.


Syntax: include:
To search a call's Type [START] and [START]welcome Contacts where the

opening for a specific then the word or word “welcome” was
word or phrase phrase of your used in the call's
interest without a opening, among the
space between them. first 100 words of the
conversation (around
30 seconds on
average).
To search a call's Type [END] and then [END]"thank you Contacts where the
closing for a specific the word or phrase of phrase “thank you”
word or phrase your interest without was used in the call's
a space between closing, among the
them. last 100 words of the
conversation (around
30 seconds on
average).
Contacts where a Inside the [START] [START:60]welcome Contacts where the

specific word or operator, specify the word “welcome” was
phrase is required number of used in the call
encountered in the words after the colon opening, among the
call's opening, among (:). first 60 words of the
a certain number of conversation (around
first words. 20 seconds on
average).
Contacts where a Inside the [END] [END:60]"thank you Contacts where the
specific word or operator, specify the phrase “thank you”
phrase is required number of was used in the call's
encountered in the words after the colon closing, among the
call's closing, among (:). last 60 words of the
a certain number of conversation (around
last words. 20 seconds on
average).

Unauthorized Characters
The following characters cannot be used:
 Underscore (_)
 Equal sign (=)
 Square brackets ([]) - except for cases when they are used as part of the [START]
and [END] operators.
 Backslash (\)
 Arrow heads (<>)
 Question mark (?)
 Asterisk (*)
 Pound (#)
 Operators AND, NOTIN, NEAR, [START], and [END] must be entered in upper case
only. If you type any of these operators in lower case (for example, and instead of
AND), the Speech Analytics application searches for them as for keywords.
 NOT, if typed in capital letters, is regarded as an operator. Searches with the NOT
operator retrieve results where one term is more dominant than the other.
For example, “bill NOT credit” will retrieve contact segments where “bill” is more
dominant than “credit”. Contact segment containing “credit” will be assigned lower
ranks than results containing “bill” only. Contact segments where “credit” is more
dominant than “bill” will not be retrieved.
 Keywords or phrases for which you are searching are not case-sensitive.
 Avoid searching for frequently used words (for example, 'hello” or “yes”). Widely
used terms by nature do not point to specific issues; therefore contacts where these
words are found receive lower ranking, which might prevent them from appearing in
search results.
Building More Complex Text Queries

By combining operators, words and phrases, you can build more complex and specific
text queries, narrowing search results to the most relevant contact segments only. Use
the parenthesis () to group parts of your query.
The following are some examples of more complex queries.
 (call NEAR back) NOTIN “call back today”
This query will retrieve all contact segments where phases similar to “call you back”,
“call me back”, “didn't not call me back” were detected, excluding the segments
where the phrase “call back today” was found.
 “called before” “called yesterday” “called earlier”
This query will retrieve contact segments where one or more of these phases were
used: “called before”, “called yesterday”, or “called earlier”.
 [END:60]C:”thank you”

Chapter 4 - Contact Search and Analysis Guided Search
With such a text query, the Speech Analytics application will retrieve contact
segments where customers used the phrase “thank you” in the call's closing, among
the last 60 words (around 20 seconds on average) of their part of the conversation.
 [START](how NEAR can NEAR help)
This query will retrieve contact segments where the phrase “How can I help you?”
or similar phrases (for example, “How can we help you?”) was used in the call's
opening, among the first 100 words of the conversation (around 30 seconds on
average).
 [START]close NEAR account
[START](close NEAR account)
These two queries will retrieve the same results - all contact segments containing
phrases such as “close my account”, “close your account”, “close the account”,
“close my recent account” and similar in the call's opening, among the first 100
words of the conversation.
You cannot use parenthesis after Speaker Separation operators (A: or C:). If you want
to combine other operators with A: and C:, place the Speaker Separation operators
before every operand. For example, a query which reads A:(close NEAR account) is not
valid. You can replace this query with A:close NEAR A:account.
If you want to use the plus (+) or minus (-) sign before the [START] or [END]
operator, you must enclose the operator in parenthesis. For example,
+[START:50]account is not a valid expression. Instead, you should write
+([START:50]account).
Guided Search
The Guided Search assists you in performing keyword searches in Speech Analytics.
This section outlines the information required for understanding guided searches, by
 Using Autocompletion in Text Queries, page 77
 Context-Based Suggestions, page 80
 Working with Context-Based Suggestions in a List, page 81
 Working with Context-Based Suggestions in a Tree, page 83
 Understanding the Term Tree, page 84
Using Autocompletion in Text Queries

Autocompletion is activated every time you type the first three characters (unless
configured differently) of a word in your text query - either when performing keyword
searches or when defining category terms. With autocompletion, Speech Analytics starts
searching as you type and automatically displays a list of possible word completions,
from which you can select words for your query. The list is compiled from the words that
appear in transcribed contact segments.

The following is an example of the autocompletion list for a keyword search. The
example shows completion options that were displayed after “pass” was typed in the
Search field.
In addition to ensuring correct word spelling and decreasing the amount of time spent
typing words, autocompletion also performs a number of useful functions:
 Word Grouping, page 78
 Word Frequency and Sorting, page 78
 Selecting Words, page 79
 Closing the Autocompletion List, page 79
 Viewing Options, page 79
 Searching with Operators, page 80
 Disabling Autocompletion, page 80
Word Grouping
To help you identify related words, all items in the autocompletion list are grouped by
their stem, including the stem variations. For example, the “broke” group can include
not only words of the same stem (“broken”, “broker”) but also the stem's variants
(“break”, “breaks”, “breaking”), provided these words were detected in transcribed
contact segments.
Word Frequency and Sorting

Next to each word, you can see the number of contact segments in which this word was
found (regardless of the number of occurrences or scores of the word in any of the
segments), and their percentage out of the total number of transcribed contact
segments.

The most common word in each group appears twice:

 As a base word of the group - to allow selection of all words of the group.
 At the top of the list inside the group - to enable you to select that word only.
Word groups are sorted by frequency of their base word, with the group of the most
commonly used word at the top. Words within each group are sorted by frequency too.
In the example above, the “password” group appears at the top of the list, which means
that this is the most commonly used word. From the autocompletion list, you can see
that this word was found in 350 segments, which constitutes 6.71% of all transcribed
segments. The word “password” serves as the base word of this group and also appears
in the list inside the group. Such words as “pass” and “passenger” were less frequently
used in transcribed contacts, so their groups appear further down in the list.
NOTE Search results for a selected word may include fewer contact segments than
indicated in the autocompletion list. This happens because the Speech Analytics
application excludes results with lower than two-star ranks from the Results page.
A similar exclusion occurs in the Category Builder, where contact segments with
ranking lower than 3 stars (or as defined in the category) are not displayed.
Selecting Words
 Click any single word in the autocompletion list to add this word to your definition.
 When performing keyword searches, click the base word of an expanded group to
add all words of the group to the query. If you need to add the base word only,
select it from the list within the group, or click it when the group is collapsed.
NOTE Word group selection is not available for category term definitions in the Category
Builder.
Closing the Autocompletion List

 The list disappears as soon as you select a word from it.
 If you wish to close the autocompletion list without selecting any words from it, click
anywhere outside the autocompletion list, or press the Esc button.
Viewing Options
Make your work with the autocompletion list more comfortable:
 To view the entire list of terms, use the scroll bar on the right side of the list, or
resize the window by dragging the lower right corner of the window to the required
size.
 Collapse any word group in the list by clicking the minus sings (-) to the left of the
base word of the group.

Searching with Operators

When searching with operators (for example, NEAR, A:, C:, [START], and [END]),
autocompletion recognizes the operators and ignores them. Completion is only provided
for the characters that follow the operator.
If you are searching using the Speaker Separation operators, A: or C:, the header of the
autocompletion list reads “Word occurrences in the agent's part” or “Word occurrences
in the customer's part” respectively. In the example above, no Speaker Separation
operators were used, so the header states “All occurrences of the word”.
For more information, see “Building Text Queries” on page 70.
Disabling Autocompletion
Autocompletion is always enabled in the Search field, but you can disable it in the
Category Builder.
The change applies to the category definition you are working on and is effective during
your current working session only. The next time you open the Category Builder,
autocompletion is enabled in all category definitions.
To disable autocompletion:
In the Category Builder, in the category definition page, clear the Allow term auto
completion check box in the My Term area.
For more information on the category definition page, see “Understanding the Category
Definition Page” on page 168.
Context-Based Suggestions
The Speech Analytics application assists you during your search by offering context-
based suggestions for your text queries. The suggestions give you ideas on other related
terms to search for and help identify contexts in which these terms have been used.
When preparing a list of context-based suggestions, the Speech Analytics application
first compiles a list of all transcribed contact segments that contain the term. It then
uses a special algorithm to analyze the words and phrases that appear in these
segments and displays up to 50 of the most statistically significant terms, including
phrases (appearing in quotation marks). These are the terms that have the strongest
correlation with the term of your text query.
If you are performing a keyword search, the suggestions are based on the entire volume
of transcribed contacts. In the Category Builder, the scope of segments that the Speech
Analytics takes into account for preparing the list depends on the task you are
performing. If you click the Suggest for Term button, the list is the same as in a
search. When you click the Suggest for Category button, the Speech Analytics takes
into account only those contacts that meet the category definition.
If your text query includes such operators as A:, C:, [START], and [END], the suggested
items are correlated accordingly. For example, if you use the A: operator, the
suggestions are based on the agent's part of the calls. For more information, see
“Building Text Queries” on page 70.

If the text query contains more than one term, suggestions refer to any of the terms.
This section outlines the information required for understanding context-based
suggestions for your text queries by covering the following topics:
 Context-Based Suggestions: List versus Tree View, page 81
Context-Based Suggestions: List versus Tree View

You can view context-based suggestions for your terms in a list or in a tree view. In both
cases, the same terms are displayed, but different additional information is provided.
The list shows correlation of each suggestion with your search term, while the tree
enables you to see how the suggested terms are correlated with each other and how
they are distributed among different contexts.
Learn about each way of viewing context-based suggestions in the following topics:
NOTE You cannot view context-based suggestions in a list and in a tree simultaneously.
Working with Context-Based Suggestions in a List

1 Perform one of the following:
 Type the term into the Search field and click More Suggestions... to the right
of the Search field.
A list of suggestions based on the entire volume of transcribed contacts appears.

 When defining category terms, enter a term in the My Term field and click the
Suggest for Term button, or click the Suggest for Category button.
A list of suggestions, which is based on the transcribed contacts that meet the
category definition, appears.
The blue bar next to each suggestion serves as a visual indication of the
suggestion's correlation with your search term.
If Speech Analytics detects related terms among suggestions, it groups them under
the term of the highest correlation. The groups are automatically expanded.
The icon to left of a “child” term indicates its relationship with the “parent” term and
can be any of the following:
 - a grammatical variant of the suggested term.
 - a synonym of the suggested term.
 - a phrase with the exact occurrence of the suggested term.
The list is sorted by correlation, with more correlated suggestions displayed at the
top.
When a term group is expanded, relevance bars appear for each of the “child”
terms, but for the “parent” term. To view the correlation bar of a “parent” term,
collapse the group by clicking the minus sign (-) to the left of the group. Notice that
the relevance of the “child” terms is lesser than of its “parent”.
TIP To view the entire list, use the scroll bar to the right of the list or resize the window
by dragging its lower right corner. Resizing is not available in the Category Builder.

2 Review the suggestions and add the relevant ones to your query:
 When performing a search, click, one by one, all the terms you want to add to
your query. Each term that you click is added to the Search field, and becomes
bold in the list. If you wish to remove a suggestion that you just added, click the
suggestion in the list again. For more information on performing keyword
searches, see “Searching by Keywords or Phrases” on page 67.
 When defining a category term, select a single term from the list. The selected
term is displayed in the My Term field, replacing the previously displayed term.
For more information on defining category terms, see “Defining Category Terms
in Classic View” on page 182.
 Clicking a “parent” term adds this term only to your query. If you need to select
any “child” term of the group, select that specific term.
3 To close the list of suggestions when performing a search, click anywhere outside
the list, or click the Close button in the top right corner of the window.
Working with Context-Based Suggestions in a Tree

In addition to viewing context-based suggestions for your search term in a list, as
described in “Working with Context-Based Suggestions in a List” on page 81, you can
display the same suggestions in a tree.
The tree enables you to see how the suggested terms are related to one another and in
which contexts they have been used. To learn about the term tree, see “Understanding
the Term Tree” on page 84.
To view and select contact-based suggestions in a tree:

1 Type the term in the Search field or create a Filter.
2 Click the Visualize button to the right of the Search field.
The initial Visualize window appears, showing the suggested terms in a tree. Your
search term appears in a bold dark-grey font, while the rest of the terms appear in
red.
3 Review the tree to understand which terms are more correlated with each other,
and the contexts in which the terms have been used.
4 To see the entire window, click the Maximize button in the top right corner of the
window to enlarge the window.
5 If you want to change the search by adding or removing terms to the current
search, click the terms in the tree.
Each term that you click is added to the Search field, and becomes bold in the tree.
To remove the term from the Search field, click the term again.

6 Click Visualize to see the results for your new search request in the Visualize
window.
Clicking Find in the Visualize window after you entered or removed words from the
Search text box, closes the Visualize window and returns you to a list of contacts
generated according to your Find request in the Contact Analysis tab.
Once you click the Find or Visualize button in the initial Visualize window, the
Find button is replaced by the Find Within Results and New Find buttons.
The Find Within Results button will preform a search within your last search
results. The New Find button will preform a new search. Clicking both buttons
returns you the Contact Analysis tab.
7 To close the Visualize window, click the Close button in the top right corner of
the window.
Understanding the Term Tree

A term tree is a complex structure (see “Understanding the Term Tree Structure” on
page 86) that shows you how terms are correlated with each other and how they are
distributed among different contexts.
The term tree is displayed when you visualize context-based suggestions for search
terms. You can also view the term tree when defining terms for your categories in the
Graphical View. In both cases the main elements of the tree are the same, but the
Graphical View also provides additional information for each term. To learn more about
the specifics of the terms tree in the Graphical View, see “Understanding the Graphical
The following is an example of a term tree in the Visualize window.

The main elements of the term tree are:

 Nodes, page 85
 Terms, page 85
 Branches, page 86
Nodes
There are three types of nodes in the term tree: leaf nodes, middle nodes, and one
central node.
 Leaf nodes are final nodes on each branch that display terms.
 Middle nodes do not contain terms and serve as branching points only.
 The central node serves as a starting branching point of the tree. See
“Understanding the Term Tree” on page 84 for more details on the central node.
Terms
On each leaf node, you can see a group of terms that were clustered together because
of strong correlation in their usage.

The groups are half-transparent, so that you can see to which branch each group
belongs.
If groups are overlapping each other, point to one of the groups with your mouse to view
this term group in a clearer way, as shown in the picture below.
Overlapping Groups (No Selection)
Overlapping Groups with One Group Selected
Branches
Typically, each branch in the term tree represents a separate context. Term groups that
appear on the same branch are closely related.
NOTE Term groups that appear on different branches are not necessarily unrelated.
Understanding the Term Tree Structure

The term tree has a hierarchical structure where branching starts from one root node
and continues further down to leaf nodes. In the term tree, the layout of this hierarchy
is determined by contexts in which terms have been used. The root node becomes a

central branching point, from which branches spread in different directions, whereas
branches of related contexts are positioned closer to each other.
TIP Locating the central node can be helpful in understanding the structure of
the term tree. You can differentiate the central node by two lines that lead
from it to its child nodes. Unlike the central mode, each middle node has
three leading lines - one to its parent node, and two lines to each of their
child nodes.
To better understand the layout of the tree, let's take a close look at some examples.
Example 1
Term groups A and B belong to one context, and groups C and D belong to a different
context.
In this case, groups A and B will be placed on one branch in the tree, and groups C and
D will appear on a different branch. A traditional presentation of this structure is shown
below.

To better show the two different contexts in which the terms are used, the same
structure will be presented differently in the tree view.
The following is an example of the term tree with a similar structure.
The tree has two distinct branches related to different contexts: the upper branch refers
to billing issues, the lower branch refers to a wider context, defined by such terms as
“date”, “data”, “information”, “ask” and “access”.

Example 2
Term groups A, B, and C belong to one context, with a closer relation between groups A
and B; group D belongs to a different context.
In the tree, groups A, B and C will be placed on one branch in the tree. This branch will
have two sub-branches - for groups A and B, and another one for group C. Group D will
appear on a separate branch.
The traditional view of the tree will be:
Speech Analytics will change this layout so that you can immediately notice one context
which is not related to the rest of the contexts.

The following is an example of implementing this structure in the term tree.
Terms that belong to a separate context (dollars, dollar, pay, ...) are places on a
different branch, remote from other branches dedicated to related contexts.
NOTE The examples show simplified trees. Typically, terms trees include up to 50 terms
and have more complex structures.

Chapter 4 - Contact Search and Analysis Filtering
Filtering
Using Speech Analytics filters help you focus and search for a specific type of contact.
For example, using a filter enables you to find contacts about a specific business issue, a
specific group of people, contacts recorded on a specific date, etc.
The following sections describe how to work with filters, by covering the following
topics:
 Filtering Guidelines, page 91
 Editing an Uploaded External Data File, page 94
 Deleting an Uploaded External Data File, page 94
 Setting Filters from the Results Page, page 94
 Filter List Parameters, page 95
 Checking if Search Filters are Set, page 104
Filtering Guidelines
When creating filters the following factors must be considered:
 Filters can be set from the Homepage or from the Results page. Once a filter is
defined, it applies to both pages and remains active until it is reset by clicking the
Reset Filters button.
Clicking the Reset Filters button returns the filter to its default state. The default
state is a configurable number (that is, the last number of days). This number is
configured in the Enterprise Manager.
For additional information, see “Defining Search Filters” on page 93 and “Filter List
Parameters” on page 95.
 Filter parameters can be set by manually entering one or more String values,
separated by a comma or semicolon.
 Manually entered String values can use up to 10 Wildcard characters (*) in a single
parameter field but cannot exceed 1024 characters.
 You cannot use the Wildcard character when manually entering Integer values.
 When creating a filter using an external file, consider the following:
 Filter parameters can be set with an external file that contains up to 500,000
multiple metadata values.
 The list of metadata values will be treated as filter values for the associated
metadata field within the Speech Analytics filter.
 The external file must be either a .csv file or a text file.
 The values in an external file must be separated by a carriage return (Enter).
You can also enter a space before and after each value.
 Filtering with multiple metadata values enables you to define a filter for a list of
String or Integer fields. You can filter all contacts where the associated

metadata value matches one of the values supplied in the list of String or
Integer values.
 A filter field can contain up to three external data files.
 The name of an external data file external data file is subject to Windows naming
conventions.
 Wildcard characters cannot be used in an external data file.
 You can upload an external data file or manually enter more than one value, for
each of the following parameters:
- Dialed From (ANI)
- Dialed To (DNIS)
- Custom Data 1 to Custom Data 75
- Extension
- PBX ID
- Switch Call ID
- SID Key
 You can perform recurring searches using the same filter. For more information, see
“Working with Saved Searches” on page 122.
 File List Filtering is not supported for filters associated with category definitions.
 You can check if search filters have been set and also preview the filters. For more
information, see “Checking if Search Filters are Set” on page 104.

Defining Search Filters

1 In the Speech Analytics home page, click the Set Filters button.
The Set Filter window displays the last filter set.
2 Define the parameters you want to use.
 If you want to filter according to multiple metadata values:
i. Create an external data file.
For information about uploading external files, see “Filtering Guidelines” on
page 91.
ii. Click the Upload button ( ) next to the parameter for which you want to
upload a file.
iii. In the File List Upload, click Browse and locate the file containing the
multiple metadata values.
iv. In the File Label field, enter a name for the file you selected.
v. Click OK
Once the external data file is uploaded, the parameter field will include the
file label and the data and time at which you uploaded the file.
 If you want to manually enter one or more values, simply type the values in the
relevant field (separated by commas or semicolons).
For a detailed list of all the filter parameters, see the “Filter List Parameters” on
page 95.
3 Click OK.
The Results page appears displaying a list of contact segments matching the filters
you set.
For further details about results, see “Understanding the Results Page” on
page 107.
All the charts are updated to display the number of contact segments that match the
filter that you defined.
For additional information, see:
 Editing an Uploaded External Data File, page 94
 Deleting an Uploaded External Data File, page 94
 Setting Filters from the Results Page, page 94

Editing an Uploaded External Data File

2 Click Download ( ), next to the parameter for which you want to download the
external file.
3 Save the file in a location of your choice.
4 Edit the external file.
5 Click the Upload button ( ) next to the parameter for which you want to upload a
file.
6 In the File List Upload, click Browse and locate the file containing the multiple
metadata values.
7 In the File Label field, enter a name for the file you selected.
8 Click OK.
NOTE An external file will not be updated in the filter if it is updated in the location it is
saved.
Deleting an Uploaded External Data File

2 Click Delete ( ), next to the parameter for which you want to delete the external
file.
Setting Filters from the Results Page

When you set filters from the Results page, Speech Analytics uses these filters to
present new search results based on the keywords you entered in your last query. In
some cases setting filters will refine the current search results. For example, let's say
you perform a query for the word “login” and you receive 200 results. Let's say you now
set a filter to define a date range. You should now receive less than 200 results. If you
were to now set another filter such as the agent who participated in the call you would
receive even fewer results. This is because the Speech Analytics application looks for the
contacts containing the word “login” that were recorded between a specific date range
and include the selected agent.
However, when using filters to set different date ranges, you may receive more results
with one date range than with another. For example, let's say you perform a query for
the word “login” and you then set filters from the Results page to retrieve contacts from
a specific date range. If you were to then set a different date range filter, the Speech
Analytics application will look for all contacts containing the word “login” that were
recorded during the new date range. This time you may receive more results than the
previous date range. From this example you can see that setting filters does not

necessarily limit the current results - instead it focuses the results on a specific contact
profile based on the keywords already defined.
To set filters from the Results page, click the Set Filters button, set your filters and click
OK. Click here to view a description of the available filters. Note that the Results page is
immediately updated to reflect the filters that you selected.
NOTE The filters you set in the Results page remain selected whether you return to the
home page or if you continue searching from the Results page and can also be
saved for future use. See “Working with Saved Searches” on page 122.
Filter List Parameters

Search and Filters can be based on the following properties of contact segments:
 Agent Data, page 95
 Date Range, page 97
 Contact Data, page 98
 Speaker Separation Data, page 100
 Advanced Data, page 101
 Context, page 103
Agent Data
Groups
Description:
Use the Groups search criteria to search for calls made by all agents from the selected
group(s) or subgroups, including contacts made by inactive agents.
Parameter:
To search according to groups, select the group from the Available groups box, and
click the Add Selection button to move the selected group to the Selected groups
list. You can only set filters for groups to which you were assigned, and their subgroups.
Note that you should never select a group from the Groups section and then select
agents who do not belong to this group from the Agents section. This is because the
Speech Analytics application first checks whether the agents belong to the selected
group before searching for the contacts, and if they do not, you will receive no results.
For this reason, you should either search for groups, or for agents, but there is no need
to select items from both the Groups and Agents sections.
Agents
Description:
Use the Agents search criteria to search for calls made by specific agents (rather than all
agents in a specific group).

To locate agents, select the group from the Available Groups area to which the agent
is assigned. If you are not sure which exact group the agent is assigned to, select a
group higher up in the tree. All agents that belong to this group, or to any groups that
reside below the selected group, will be displayed in the Available agents list.
Note that the Available Agents list can display up to 220 agents. If the agent you are
searching for does not appear in the list, you can search for the agent by entering all or
part of their name (proceeded or followed by a wildcard *) into the text search field
above the tree and clicking the Find button . Your system will search for the agent
within the group or subgroup you selected in the Available groups area.
Parameter:
You can also use the search feature to search for agents within the list that is displayed.
Once you have located the agent(s) you want to search for, select them in the Available
agents list and click Add Selection button to move the selected agents to the
Selected agents list. You can select up to 220 agents.
You can only set filters for agents associated with groups to which you were assigned,
and their subgroups.
Note that if you are interested in contacts made by specific agents, there is no need to
select the agents' group(s) from the Groups section.
If you select a group from the Groups section and then select agents from the Agents
section, your system will only select calls made by the selected agents if they are
assigned to the selected group(s) or their subgroups. For example, it looks for calls
made by the selected agent, provided the agents are assigned to the selected group.
Therefore, when selecting agents whose calls you want to receive, there is no need to
also select groups.
In addition, you should never select a group from the Groups section to which the
agent does not belong. This is because the Speech Analytics application first checks
whether the agent belongs to the selected group before searching for the agents' calls,
and if the agent does not belong to the group, you will receive no results.
For this reason you should either search for groups or for agents, but there is no need to
select items from both the Groups and Agents sections.

Date Range
From the last
Description:
When searching for contacts, select this option to display today's contacts plus contacts
from the last n days (where n is the number that you enter).
NOTE If you select From the last and do not change the default number of days before
Saving, the Date Range will not be part of the category definition.
If the Date Range must be part of the category definition, you must modify this
field to be a number other than the default value before saving the category.
Parameter:
From the last X days, enter how many days back to retrieve contacts from the current
day (up to 999 days). The data that is retrieved always includes today's data, so
entering 1 retrieves contacts for today and yesterday, entering 2 retrieves contacts for
today and yesterday and the previous day, etc. The From the last field does not
support 0 option, and you cannot use it to search for today's contacts only.
The search retrieves all contacts of each entire previous day included in the query plus
today's contacts.
Contacts are retrieved according to their local start time. For example, according to the
agent's local time when the contact was recorded.
Between these dates of times
Description:
Define the dates and times of the contacts you are searching for.
Click the Calendar button to select the from and to dates from a calendar.
Parameter:
This field is selected by default when performing keyword searches and when selecting a
category, from the Categories area. By default, the date range filters all available
contacts in the Speech Analytics application from the last default number of days.
This is not available when defining filters for categories.
Note that contacts are retrieved according to their local start time, (for example,
according to the agent's local time when the contact was recorded). Therefore, if you
are on the West Coast of the US and the local time is 11 AM, and you are searching for
contacts that occurred between 7 AM and 11 AM, you will not receive contacts from the
East Coast that occurred in the last three hours as the local time there is 2 PM, however,
you will receive calls that occurred between 7 AM and 11 AM East Coast local time.
All Dates
Description:
Select to show all the contacts.
Parameter:

When this option is selected, your search results can contain all the contacts in the
Speech Analytics index.
Time of day
Description:
Select a specific time range from the list or leave the Any option to look for contacts that
were recorded throughout the day.
Parameter:
The definition of each of the options, Morning, Noon, Afternoon, Evening or Night, is
defined in the Enterprise Manager.
Contact Data
The Contact Data area of the filter page enables you to create filters based on basic
contact properties.
Name: Description
Duration Enter the minimum and/or maximum contact duration in the

seconds:miliseconds format. The minimum and maximum field
can contain only numeric characters.
Dialed from (ANI) Enter the Dialed from (ANI) number for which you want to
search. The value can contain up to 15 alphanumeric
characters.
The Dialed from (ANI) field can contain one or more values, by
either manually entering the values (separated by a comma or
semicolon) or by uploading a file containing the required
metadata values.
Dialed to (DNIS) Enter the dialed to (DNIS) number for which you want to
search. The value can contain up to 15 alphanumeric
characters.
The Dialed to (DNIS) field can contain one or more values, by
metadata values.
Direction Incoming to search for inbound contacts

Outgoing to search for outbound contacts
Internal to search for contacts performed within the contact
center
Unknown to search for contacts for which the system cannot
determine the contact direction.
<Any> to search for any of the above contact directions.

Name: Description
Data Set The Data Set field is displayed only if your system supports
Custom Data sets.
The field is a drop-down list displaying existing custom data
sets. Select a value from the list to view custom data fields
that are relevant for the set. The fields appear in the order
that have been defined for this set in the CD Manager.
Note that you can view only those fields that have been
assigned to you in the Assignment Manager application.
The default value in the Data Set field is either the first set in
the list, or the one which is most appropriate for the current
search.
Custom Data Fields The fields that appear are those that were assigned to you in
the Assignment Manager application. Each contact can contain
up to 30 contact data fields. If you try to select more than 30,
an error message will appear.
This field is case-sensitive.
If the contact custom data field is a text box, enter the value
for which you wish to search. You can enter up to 1024
characters. You can filter according to a word in a data field by
using the * wildcard. For example, entering Sal* will retrieve
both the words Sale and Sales. Note that the wildcard search
is case-sensitive. If you enter S*, the search will retrieve
words that begin with a capital S.
If the From and To fields appear, enter a range of numbers.
If a list box appears, select a value from the list. The values
that appear in the list are those that were assigned to you in
the Assignment Manager application. However, if you were
also assigned the right to “Search for any value”, you can use
the Any option to search for calls belonging to all five lines of
business.
If the <All> option appears, select it to search for all the
values that appear in the list.
If the <Any> option appears in the list, select to search for all
contacts, regardless of their custom data, including those
custom data fields that were not assigned to you and thus do
not appear in the list.
The Custom Data field can contain one or more values, by
metadata values.
NOTE For a detailed description of each of these properties, see “Understanding Contact
Details” on page 116.

Speaker Separation Data

The Speaker Separation Data area of the filter page is displayed only if Speaker
Separation has been enabled in your system. This area enables you to create filters
based on data that is available only for those contact segments recorded with Speaker
Separation.
NOTE These tables describes how to use these properties for filtering. For a detailed
description of each of these properties, see “Understanding Contact Details” on
page 116.
Name: Description
Agent Talk-Time Enter the minimum and/or maximum percentage of the agent
Percentage talk time in a contact segment to search for. Use integer
values in the range from 0 to 100.
Customer Talk-Time Enter the minimum and/or maximum percentage of customer

Percentage talk-time in a contact segment to search for. Use integer
values in the range from 0 to 100.
Talk-over Time Enter the minimum and/or maximum percentage of talk-over

Percentage time in a contact segment to search for. Use integer values in
the range from 0 to 100.
Number of Agent Enter the minimum and/or maximum number of times the
Initiated Talk-overs agent interrupts the customer during the same contact
segment.
Number of Customer Enter the minimum and/or maximum number of times the
Initiated Talk-overs customer interrupts the agent during the same contact
segment.

Advanced Data
The Advanced area of the filter page enables you to create filters based on Enhanced
Filters.
Name: Description
Exception In the Exception field, select:

<Any> to search for any contact segments, regardless of
their belonging to exceptional contacts
Only Exception Contacts to search for segments that belong
to contacts that qualify as exceptions
Non-exception Contacts to search for segments that did not
belong to contacts that qualify as exceptions.
Extension Enter the extension(s) for which you want to search.

The Extension field can contain one or more values, by either
manually entering the values (separated by a comma or
metadata values.
PBX ID Enter the PBX ID(s) for which you want to search. If the
switch/CTI Server does not provide PBX ID, enter the login
name in this field.
You can search for a PBX ID using the * wildcard. For example,
entering 22* will retrieve contacts with both agent 2244 and
agent 2255.
The PBX ID can contain up to 1024 alphanumeric characters.
The PBX ID field can contain one or more values, by either
metadata values.
Number of Holds Enter the number of holds for which to perform the search.
Number of Transfers Enter the number of transfers for which to perform the search.
Number of Conferences Enter the number of conferences for which to perform the
search.
Total Hold Time Enter the minimum and/or maximum time in seconds or
seconds and miliseconds the customer was put on hold in the
following format seconds:miliseconds.
Wrap-up Time Enter the minimum and/or maximum time in seconds or

seconds and miliseconds that the agent took to wrap up the
call, in the following format seconds:miliseconds.
Switch Select the switch to which the agent logs in.

Name: Description
Switch Call ID The Switch Call ID field can contain up to 1024 alphanumeric
characters. You can search for a word in a data field by using
the * wildcard. For example, entering Sal* will retrieve both
the words Sale and Sales. Note that the wildcard search is
case-sensitive. If you enter S*, the search will retrieve words
that begin with a capital S.
The Switch Call ID field can contain one or more values, by
metadata values.
SID Key The SID Key field can contain one or more SID Keys, by either
metadata values.
Silence Time Enter the minimum and/or maximum percentage of silence

Percentage time in a contact segment to search for. Use integer values in
the range from 0 to 100.
NOTE For a detailed description of each of these properties, see “Understanding Contact

Context
Name Description
Speech Analytics From the Available Categories select the categories that you
Application Category want to search for and click the Add Selection button to move
the categories to the Selected Categories list.
If you select multiple categories, the Speech Analytics application
will search for contacts that belong to any of the selected
categories.
Note that your system will look for contacts that belong to any of
the selected categories. If you want to find contacts that belong to
two categories, select one of the categories from the Available
Categories field, and then click on the second category in the
Speech Analytics application home page. For example, if you are
interested in all problems concerning a new online service that you
have launched, you can select the Online Services category as a
filter, and then you can select Customer Complaints from the
Categories list. This will find all contacts that refer to the online
service but are also customer complaints.
Alternatively, you can click on the specific category from the
Category chart in the home page, and then click on the second
category from the Category chart in the Results page.
Emotions Select <Any> to search for all contacts, regardless of the presence
of emotions.
Select Contacts displaying Emotions to retrieve only contacts
where emotional words were used.
Categorization Select <Any> to search for all contacts, regardless of the

categorization.
Select Categorized Contacts to search for calls pertaining to a
specific business issue among the total scope of contacts, such as
customer complaints, billing issues, product feedback, or repeat
calls.
Select Uncategorized Contacts to search for contacts that do not
meet any of the existing category definitions. Uncategorized
contacts allows you to search for and analyze (using existing
contact details), contacts in the Speech Analytics index that are not
assigned to any category.

Chapter 4 - Contact Search and Analysis Using Categories to Find Contacts
Checking if Search Filters are Set

If a filter is set, the bar above the home page and Results page includes a Filter icon
.
 Hover your mouse over the Filter icon above the home page or Results page to view
the filters in a tooltip.
In the Results page, you can also use the History bar to check if any filters have been
set. If any filters have been set, the History bar includes the Filters set entry.
 Hover your mouse over Filters set entry in the History bar of the Results page.
TIP To be able to edit the filter, click the Set Filters button. For more information, see
“Defining Search Filters” on page 93.
Using Categories to Find Contacts

Searching by a category retrieves all contacts that belong to that specific category.
These are published and active categories only; inactive categories are not used for
contact categorization.
To find contact segments belonging to a specific category:
 In the Speech Analytics home page, in the Categories area, click the category
name.
or
Click the category's bar in the Category chart in the lower right corner of the home
page.
The Speech Analytics Results page appears displaying all the category's contact
segments that you are authorized to view. The minimal rank of the found contact
segments is predetermined by the category's rank threshold. For example, if a
category's rank threshold is four stars, then the result set will include segments with
four- and five-star ranks only. For more details on the ranking of segments in a
category, see “Defining the Rank Threshold of a Category” on page 195. Note that
all results in the Results page are according to the custom data and conditional
custom data fields that were assigned to you in the Assignment Manager.
For further details about results, see “Understanding the Results Page” on page 107.
You can also retrieve contacts belonging to a specific category by clicking the relevant
bar in the Category chart in the Results page. In this case, the search will be executed
within found results.
NOTE Category searches use the Inverse Document Frequency (IDF) weight when
calculating search results, to find the relevancy of all terms in categories. For more
information on IDF and the way Speech Analytics calculates search results, see
Ranking and Document Relevancy, page 223.

Chapter 4 - Contact Search and Analysis Using Categories to Find Contacts

Chapter 4 - Contact Search and Analysis Searching for Contacts that Indicate Emotion and Dissatisfaction
Searching for Contacts that Indicate Emotion

and Dissatisfaction
The Speech Analytics application includes a special emotion detection feature which
enables you to retrieve a list of all the contacts that indicate customer dissatisfaction or
any other kind of emotion.
NOTE The definition of what is considered an indication of emotion is customizable, to

suit each contact center's needs, and the Emotions category is defined when your
system is set up. Like other categories, the Emotions category can comprise not
only terms but also filters and a dollar value.
To retrieve a list of all contacts indicating dissatisfaction or emotion:

 Click the Emotions category in the Speech Analytics home page. The
Results page appears displaying a list of contacts matching the search criteria.
For further details about results, see “Understanding the Results Page” on
page 107.
The Results list will only display contacts associated with groups (and their
subgroups) that were assigned to you in the Assignment Manager.
The Results list will only display contacts associated with Contact Data and
Customizable Contact Data values that were assigned to you in the Assignment
Manager.
Alternatively, if you want to perform a keyword or category search, but you are still
interested in seeing which contacts within your search results indicate dissatisfaction or
emotion, look for the contacts displaying the Emotions icon in the E column.
Understanding Search Results

The Results page displays a list of segments that meet your search criteria. The charts in
the lower area of the page are updated to include only those segments that belong to
the current result set.
This section outlines the information required for understanding Search Results, by
 Understanding the Results Page, page 107
 Understanding the Results List, page 109
 Viewing Collapsed Details, page 112
 Viewing Query Results Metrics, page 113
 Working with Contact Details, page 114

Chapter 4 - Contact Search and Analysis Understanding Search Results
Understanding the Results Page

The following is an example of the Speech Analytics Results page.
Results Page Legend:

 Click the Home link to display the home page in the Contact Analysis tab.
 The Help button in the top right corner of the application window allows you to
access the application's online help at any time.
 The toolbar at the top of the Results page displays buttons for paging through the
results of your search. Other buttons on the toolbar allow you to view the
TellMeWhy™ analysis for the retrieved results, set or reset filters, save your search
and create reports.
 Click the Query Result Metrics button to display query result metrics for contact
segments retrieved in the Results page. See “Viewing Query Results Metrics” on
page 113.
 The Search field allows you to perform searches by keywords or phrases, while
using such guided search options as autocompletion and context-based
suggestions. From the Results page, you can start a new search or continue
searching within found results. For more information, see “Searching in Speech
Analytics” on page 66.
 The History bar enables you to track your latest search operations and to access
search results associated with your previous search criteria. See “Using your Search
History” on page 125 to learn more about using and understanding the History bar
of the Results page.
 The banner above the list provides a summary of the results displayed on the page.
You can view the number of retrieved contact segments and their percentage out of
the total number of contacts. If search filters have been set, this area allows you to
preview the filters. For details, see “Checking if Search Filters are Set” on page 104.
 The banner above the results list provides a page navigation. The page navigation
contains two buttons that enable you to jump to the next or previous search results
page.

In addition, if a filter has been applied, the page navigation contains a filter icon.
Moving your mouse over the filter icon displays information about the applied filter.
The following image, shows both the page navigation controls and the filter icon.
 The contact list displays a detailed list of contacts that were retrieved according to
your search criteria and enables you to playback the retrieved contacts. For a
description of each of the columns in the list, see “Understanding the Results List”
on page 109.
 By default, the bottom of the page displays charts of retrieved contacts. See “Using
Charts” on page 127 to learn more about charts and ways of working with them.
If you choose to playback a contact, the charts are replaced with the Speech
Analytics player, through which you can control the contact playback. For more
details, see “Understanding the Speech Analytics Player” on page 151.
If a chart does not appear, see “Troubleshooting” on page 229.
If your search results produced more than 10,000 results, see “Troubleshooting” on
page 229 to learn how to view them.
When you select a contact for playback, the charts at the bottom of the Results page are
replaced by the playback area. see “Understanding the Speech Analytics Player” on
page 151 for a description of this area.
Alternatively, contacts can be played back in your default Quality Monitoring application.
For more details on this type of playback, see “Opening a Contact” on page 158.
From the Results page, continue working in one of the following ways:
 Understanding the Results List, page 109
 Playing Back Contacts in Speech Analytics, page 150
 Starting a New Search, page 148
 Searching within Found Results, page 121
 Viewing TellMeWhy Analysis, page 160

Understanding the Results List

When looking at the Results list, note the following:
 The Results list will only display contacts associated with groups (and their
subgroups) that were assigned to you in the Assignment Manager.
 The Results list will only display contacts associated with Contact Data and
Customizable Contact Data values that were assigned to you in the Assignment
Manager.
The Results List is made up of the following parameters:
 No., page 109
 Checkbox, page 109
 Rank, page 109
 Play this Contact, page 110
 Keywords Found, page 110
 E, page 111
 Categories, page 111
 Agent, page 111
 Local Start Time, page 111
 Screen Exists, page 111
 Duration, page 111
 Contact Details, page 112
No.
No. is a sequential list. The contacts with the highest relevance to the search performed,
appear at the top of the list. For additional details, see “Rank” on page 109.
Checkbox
Select the checkbox to mark the contact segments that you want to include in the
contact data report. For more details, see “Checkbox” on page 109.
Rank
Ranking is the process by which a contact segment is given an overall score in a search.
The rank is composed of different factors, including a system confidence and frequency
factor, otherwise known as Inverse Document Frequency (IDF). By default, in single-
term queries, the IDF factor is not included in the calculation. For more details on IDF
and when this weight is used in searches, see Ranking and Document Relevancy,
page 223.

Rank in Keyword Searches

In keyword searches, each contact segment is ranked according to the keywords and
phrases it contains. A maximum of five stars or a minimum of two stars appears in the
Rank column when searching by keywords, with five stars given to the highest ranking
contacts. The returned contact segment list is displayed in descending order, with the
highest ranking contact segments appearing at the top of the list.
Rank in Category Searches
In category searches, each retrieved contact segment is ranked according to its
relevance for the category. A maximum of five stars can be given to the highest ranking
results. The minimal rank of the returned contact segments is predetermined by the
category's rank threshold. For example, if a category's rank threshold is four stars, then
results will include segments with four- and five-star ranks only. For more details on
ranking segments in a category, see “Defining the Rank Threshold of a Category” on
page 195.
The results are displayed in descending order, with the highest ranking contact
segments appearing at the top.
Rank in Filtered Searches and in Searches for All Contact Segments
When you performed a search with a filter only, or when searched for all contact
segments in the online database (by clicking Find in the home page), all results are
displayed with 5 stars, in chronological order with the newest contact at the top of the
list.
The Speech Analytics online database, also referred to as the Index, stores transcribed
contact segments and their details. All search, playback and analysis operations in
Speech Analytics are performed against the Speech Analytics online database. If a
contact segment is no longer available in the online database, it cannot be retrieved in
searches or played back. Speech Analytics online database stores data of a relatively
recent time period, up to a few months at most.
Play this Contact

Play this Contact enables you to launch the segment playback. For more details, see
“Playing Back Contacts in Speech Analytics” on page 150.
When you start the playback, the Play this contact button turns into a Close the
player button , and the Speech Analytics Player appears at the bottom of the Results
page, replacing the charts.
You can click the Close the player button to close the player and to display the
charts again.
Keywords Found
Keywords Found displays the keyword(s) (from the keyword search or from the selected
category) that were found in the contact and the number of times that they appeared.
If the contact contains multiple keywords and their list is too long to fit one line, the list
appears collapsed. For details on viewing the entire list, see “Viewing Collapsed Details”
on page 112.

E
The E column displays the Contact Displaying Emotions icon if the contact indicates
dissatisfaction or emotion.
The column enables you to see whether a certain contact displays more emotion/
dissatisfaction than other contacts, even before you have listened to it.
The definition of what is considered an expression of emotion is defined as part of an
Emotions category in the Category Builder application.
Categories
Categories contains the names of the categories in which the contact appears.
Even if you selected a specific category, the search result list displays all categories to
which the contact belongs. These are only published and active categories; inactive
categories are not used for contact categorization.
If the contact is associated to several categories, and their list is too long to fit one line,
the list appears collapsed. For details on viewing the entire list, see “Viewing Collapsed
Agent
Agent represents the name of the contact's agent, if known; otherwise, Unknown
appears.
Local Start Time

Local Start Time represents the contact's local start time. For example, the agent's local
time at the time of recording.
Clicking on the start time opens the contact in the Workspace page.
If you have access to more than one application, you can define which application to
playback contacts in the Portal Preferences page.
Screen Exists
The Screen Exists icon appears in this column if a screen file was recorded for this
contact. This icon will appear even if the screen file has already been deleted from the
system.
Duration
Duration represents the length of the contact.

Contact Details
Clicking the View contact details icon displays the contact details. See “Viewing
Contact Details” on page 115 for more information
Viewing Collapsed Details

Results of various operations in the Speech Analytics application are presented as
detailed tables, with one line per each entry.
If the display of the details requires more than the available one line in that table grid,
these details appear collapsed, with a plus sign (+) at the beginning and an ellipsis (...)
at the end of the cell.
You can see examples of collapsed details in the following locations:
 In the Category list, when a term list for a category contains multiple terms.
 In the Results page:
In the Categories column when a contact is associated with several categories.
In the Keywords Found column when multiple keywords have been found for a
contact.
The following is an example of a collapsed list in the Keywords Found column in the
Results page.
You can view the entire content of the collapsed cell in the tooltip or by expanding the
list.
To view the tooltip, place the mouse pointer over the ellipsis sign (...) at the end of the
cell.
The tooltip disappears as soon as you move the mouse pointer from the ellipsis sign (...)
at the end of the cell.
1 Click the plus sign (+) to expand the cell. The plus sign (+) changes to minus (-),
and the row's height is adjusted to display the entire content of the cell.
2 To collapse the cell, click the minus sign (-).

Viewing Query Results Metrics

You can view query result metrics of any result set displayed.
 In the Speech Application window, click the Query Result Metrics button .A
new window opens on top of the page, displaying the metrics of the current result
set.
When you open the application and work in the Trend Analysis tab or in the Contact
Analysis tab without performing any searches, query result metrics refer to all
transcribed segments.
When you are working in the home page, query result metrics refer to all
transcribed segments, or, if any filters have been defined, to the results that match
the filter.
If the Results page is displayed or you return to the Trend Analysis tab after
performing a search, the metrics refer to the result set in the Results page.
The following is an example of query result metrics.
The query result metrics include:

 Average Silence-Time Percentage
 Average Number of Holds
 Average Number of Transfers
 Average Duration
 Average Total Hold Time
If Speaker Separation for Speech Analytics has been enabled, an additional set of data
(Speaker Separation metrics) is provided. These details include:
 Average Agent Talk-Time Percentage
 Average Customer Talk-Time Percentage
 Average Talk-Over-Time Percentage
In systems that work in mixed mode and contain speaker-separated segments, as well
as segments recorded without Speaker Separation, search results may include both
types of segments. Speaker Separation data metrics for such results are calculated only
on speaker-separated segments. In this case, an asterisk is displayed next to each of

the Speaker Separation data metrics. Hover your mouse over the data to view a tooltip
with an explanation on its calculation.
See “Speaker Separation” on page 18 for more information on using Speaker Separation
in Speech Analytics.
See “Understanding Contact Details” on page 116 for a detailed explanation of each of
the query result metrics.
Query Result Metrics in Charts and Reports

You can also view each of the query result metrics in a dedicated chart. For example, to
view the average number of holds, display the Number of Holds chart. The average
number of holds is displayed in the top right corner of the chart. For details on working
with charts, see “Using Charts” on page 127.
Query result metrics are also included in the header of contact data and chart reports.
For more details, see “” on page 212 and “Creating Chart Reports” on page 209.
Working with Contact Details

From the Results page, you can view details of any contact segment retrieved.
This section outlines the information required for understanding Contact Details, by
 Viewing Contact Details, page 115
 Understanding Contact Details, page 116

Viewing Contact Details

1 In the search results list in the Results page, locate the segment.
2 Click the View contact details icon . Contact details are displayed in a separate
window on top of the Results page.
Contact details include various characteristics that were obtained for a contact and its
segments at the time of recording (ANI, DNIS, direction, duration and more) or
automatically assigned according to their properties (for example, conditional custom
data fields, whether the contact is exception and more).
If Speaker Separation for Speech Analytics has been enabled and the segment was
recorded with Speaker Separation, an additional set of data - Speaker Separation
Contact Details - are provided for the segment as well. Use the scroll bar to the right of
the Contact Details window to view these details.

Speaker Separation contact details include:

 Number of Agent-Initiated Talk-Overs
 Number of Customer-Initiated Talk-Overs
 Total Agent Talk-Time
 Total Customer Talk-Time
 Total Talk-Over-Time
For Total Agent Talk-Time, Total Customer Talk-Time, Total Talk-Over-Time the
percentage value is provided as well. For example, if the Total Agent Talk-Time is 6:19
(45%), then the agent was talking for 6 min 19 sec, which constitutes 45% of the total
contact segment duration.
For more information on using Speaker Separation in Speech Analytics, see “Speaker
Separation” on page 18.
See “Understanding Contact Details” on page 116 for a detailed explanation of each of
the contact details.
Understanding Contact Details

Contact details with which you work in Speech Analytics can belong either to contacts or
to contact segments. A contact is a recorded interaction between a customer and the
contact center, while a segment represents an interaction between an agent and a
customer. A contact can be composed of one or more segments, depending on the
number of agents the customer spoke to. For example, if a customer speaks to one
agent only, the contact will contain one segment. However, if a customer was
transferred from one agent to another, the contact will contain two segments.
While working in the Speech Analytics application, you can view details of any single
segment and display query result metrics for the entire set of retrieved segments. If
Speaker Separation has been enabled, an additional set of data - Speaker Separation
metrics - are provided both for single segments and for search results.

Details of any contact segments can be exported to the Excel report. Contact details also
serve as filters for contact searches and for category definitions. To learn more about
the various operations related to using contact details, refer to the following topics:
 Viewing Contact Details, page 115
 Viewing Query Results Metrics, page 113
 Creating Contact Data Reports, page 213 and Creating Chart Reports, page 209
 Defining Category Filters, page 195
The following table describes all available contact and segment characteristics which you
can view and use in the Speech Analytics application. An additional Query Result Metrics
table describes query result metrics which are the average values of specific contact and
segment properties calculated for a given set of search results.
Contact and Segmentation Details
Contact Details Description
Agent Talk-Time Segment property.

Percentage One of the Speaker Separation contact details that shows the percentage of time the
agent was talking during a segment. In contact details, this parameter is displayed in
parenthesis next to the total agent talk time.
ANI Segment property.

The number from which the contact was dialed.
Custom Data Segment property.

The actual captions of custom data fields are customer-configurable and can appear
with any label (for example Account Number, Transaction, Site, Line of Business etc.)
Some of these fields may be Conditional Custom Data fields, which contain
predefined values that are based on customized rules defined by the system
administrator.
Example: The system administrator can define a Customizable Custom Data field
called Line of Business. There may also be a rule whereby the Line of Business that is
displayed is dependent on the DNIS that is dialed by the customer.
These predefined values are also assigned to you in the Assignment Manager, in
addition to the conditional contact data field itself.
Customer Talk-Time Segment property.

Percentage One of the Speaker Separation contact details that shows the percentage of time the
customer was talking during a segment. In contact details, this parameter is
displayed in parenthesis next to the Total Customer Talk Time.
Data Set Segment property.

The Data Set field is displayed only if your system supports Custom Data sets.
Dialed from (ANI) See ANI above.
Dialed to (DNIS) See DNIS below.

Direction Contact property.

The contact direction can be one of the following:
Incoming - for inbound contacts
Outgoing - for outbound contacts
Internal - for contacts performed within the contact center
Unknown - for contacts for which the system cannot determine the contact
direction.
DNIS Segment property.

The number that was dialed to initiate the contact.
Duration Segment property.

The duration a segment.
Exception Contact property.

Displays TRUE is the segment belonging to a contact that has been qualified as an
exception and FALSE otherwise.
Contacts are automatically qualified as exceptions if they meet conditions defined in
the Acquisition Director.
Extension Segment property.

The extension of the agent who handled the contact.
Number of Agent - Segment property.

Initiated Talk-Overs One of the Speaker Separation contact details that shows the number of times the
agent interrupted the customer during a segment.
Number of Contact property.

Conferences The number of times during the contact when more than two participants were
connected to the contact.
Number of Customer Segment property.

Initiated Talk-Overs One of the Speaker Separation contact details that shows the number of times the
customer interrupted the agent during a segment.
Number of Holds Contact property.

The number of times the customer was put on hold during a contact.
Number of Transfers Contact property.

The number of times the customer was transferred from one agent to another during
a contact.
Transfers to outside parties (transfer to a party outside the PBX boundaries) are not
counted as transfers by the CTI Link application.
PBX ID Segment property.

The ID of the agent who participated in this segment of the contact. This will either
be the ID the agent uses to log into the switch, or the agent's PC login name.

Silence Time Segment property.

Percentage Shows the percentage of time when neither customer nor agent were talking during
a segment. In contact details, this parameter is displayed in parenthesis next to the
Total Silence-Time.
Switch Segment property.

The switch to which the agent logs in.
Switch Call ID Segment property.

A number created by the switch, per segment of a contact.
Talk-Over Time Segment property.

Percentage One of the Speaker Separation contact data that shows the percentage of time when
the customer and the agent were talking simultaneously during a segment. In
contact details, this parameter is displayed in parenthesis next to the Total Talk-
Over-Time.
Total Agent Talk-Time Segment property.

One of the Speaker Separation contact details that shows the amount of time when
the agent only was talking during the segment. In contact details, you can also see
Agent-Talk Time Percentage.
Total Customer Talk- Segment property.

Time One of the Speaker Separation contact data that shows the amount of time when the
customer only was talking during the segment. In contact details, you can also see
the Customer Talk Time Percentage.
SID Key Contact property.
Total hold time Contact property.

The amount of time in seconds and miliseconds, the customer spent on hold during a
contact.
Total Silence Time Segment property.

Shows the amount of time when neither customer nor agent were talking during the
segment. In contact details, you can also see the Silence Time Percentage.
Total Talk-Over-Time Segment property.

One of the Speaker Separation data metrics that shows the amount of time when the
customer and the agent were talking simultaneously during a contact segment. In
contact details, you can also see the Talk-Over-Time Percentage.
Wrap up time Segment property.

The time elapsed between the end of the current contact segment and the beginning
of the next contact segment.

Query Result Metrics
Average Agent Talk Segment property.

Time Percentage One of the Speaker Separation data metrics that shows the average percentage of
time the agent was talking during a contact segment, within the retrieved contact
segments.
Average Customer Segment property.

Talk Time Percentage One of the Speaker Separation data metrics that shows the average percentage of
time the customer was talking in during a contact segment, within the retrieved
contact segments.
Average Duration Segment property.

One f the query result metrics that shows the average duration of a segment, within
the retrieved contact segments.
Average Number of Contact property.

Holds One of the query result metrics that shows the average number of holds for contacts
to which the retrieved contact segments belong.
Average Number of Contact property.

Transfers One of the query result metrics that shows the average number of transfers for
contacts to which the retrieved contact segments belong.
Average Silence Time Segment property.

Percentage Shows the average percentage of silence during a contact segment, within the
retrieved contact segments.
Average Talk-Over Segment property.

Time Percentage One of the Speaker Separation data metrics that shows the average percentage of
time when the agent and the customer were talking at the same time during a
contact segment, within the retrieved contact segments.
Average Total Hold Contact property.

Time One of the query result metrics that shows the average of the total amount of time
the customer spent on hold during a contact, for contacts to which the retrieved
contact segments belong.

Chapter 4 - Contact Search and Analysis Searching within Found Results
Searching within Found Results

With the Results page displayed, you can further refine your current search by searching
within the results. You can search for more terms, set filters, and also combine different
search methods to get to better results. See “Refining Search Results through Different
Search Methods” on page 122 for some examples of searches that involve different
search methods.
At any stage of your search, you can consult the History bar to reminded of your search
flow and to better understand the search results. You can also use the History bar to
return to any previous results of your current search. See “Using your Search History”
on page 125 for more information.
To perform a search within results:

1 With the Results page displayed, search for contacts using any of the available
search methods:
 When performing a keyword search, click the Find Within Results button.
Otherwise, if you click the New Find button, a new search will be performed.
See “Searching by Keywords or Phrases” on page 67.
 If you do not change filters, the search will be performed using previously
defined filters if there were any. See “Defining Search Filters” on page 93.
TIP Check if search filters have been set. See “Checking if Search Filters are Set” on
page 104.
2 The search is performed within the results that were found in the previous search,
and new results are displayed.
NOTE In contrast to the behavior of single-term queries, as described in “Searching by

Keywords or Phrases” on page 67, where the system does not use IDF to calculate
search results, when refining a search for a single term by using Find Within, the
system does use IDF to calculate the results. This is because a refined search is
handled like a multiple-term search, and all multiple-term searches include the IDF
weighting when calculating search results. For more information on this
functionality, see Ranking and Document Relevancy, page 223.
TIP You can use the “TellMeWhy Analysis” on page 159 to enable Speech Analytics to
suggest new search terms.

Chapter 4 - Contact Search and Analysis Refining Search Results through Different Search Methods
Refining Search Results through Different

Search Methods
To get the best search results, you can combine different search methods. The following
are some examples of such combinations.
If you want to retrieve all contact segments related to a particular problem that
occurred in the last few days, you can refine your keyword search by defining a filter
that includes the number of days back you are searching for, or a date range.
Perform your search as follows:
1 Set the time range filter.
2 Perform a keyword search related to the problem within found results.
The search results will include only those contact segments that occurred within the
specified date range and that contained the keywords you searched for.
NOTE If you enter keywords into the Search field in the home page (without clicking the
Find button) and then set filters from the home page, the Results page displays
the search results according to the filters you set, and not according to the
keyword(s) you entered. However, the keyword(s) you entered are displayed in
the Search field in the Results page, and you can now click Find within results
to perform a search with both the keywords and the filters that you set.
Another example of combining search methods is to search for contact segments by

category and then search for specific keywords within results. For example, if you
already have a category called Customer Complaints, you can look for calls that contain
complaints specific to login issues.
Perform your search as follows:
1 From the Speech Analytics home page, search by the Customer Complaints
category.
2 Perform a keyword search for the word “login” within found results.
For more details about searching by category, see “Using Categories to Find Contacts”
on page 104.
The search results will include only those contact segments that contain both the word
“login” and one or more of the keywords defined in the Customer Complaints category.
Working with Saved Searches

Saving searches can be convenient if you have to perform reoccurring search queries for
the same search criteria and do not want to have to define the criteria manually each
time you want to perform the search.
Saving a search query using a date range, such as “From the last 7 days”, can provide
you with a handy tool for searching specific criteria in a relative date range. For

Chapter 4 - Contact Search and Analysis Working with Saved Searches
example, if you want to perform a weekly search for customer complaint calls from the
last week, you can save search criteria that contains the relevant keywords and a seven-
day date range.
Saved searches can include both keywords and filters and are performed from the
Results page.
There are two types of saved searches:
 Private Search: Can only be used, deleted and/or renamed by the user who
created it.
 Public Search: Can be used by all users within the same Speech Analytics
instance. It is visible within the user’s list of available saved searches.
A public search can only be renamed and deleted by the user that created it and by
a user with Speech Analytics and Assignment Manager permissions. By default, the
Administration Group is created with these permissions.
When working with Saved Searches you can:
 Accessing Saved Searches, page 123
 Saving a Search, page 124
 Renaming a Saved Search, page 124
 Changing a Private Saved Search to a Public Saved Search, page 125
 Deleting a Saved Search, page 125
Accessing Saved Searches

Once you've saved your search criteria, you can access your searches from the My
Searches area in the Speech Analytics home page.
When you access a saved search, the search is automatically performed using the
predefined search criteria you saved.
Even if your saved search already includes filters, you can use additional filters in
combination with a saved search to further refine your results. For more information on
Filters see “Defining Search Filters” on page 93.
NOTE If you return to the home page after using a saved search containing filters, the
filters will be set. Use the Reset Filters button if you do not want to use these
filters.
To access a saved search:

1 From the Speech Analytics home page, in the My Searches area, click the name of
the search you want to view.
The system performs the search and returns a Results List, based on your selection.
NOTE A Private search is represented by the icon: and a Public search is

represented by the icon: .
For more information on pubic and private searches, see “Working with Saved
Searches” on page 122.

Chapter 4 - Contact Search and Analysis Working with Saved Searches
In some cases, your saved search may become inaccessible. This can occur if:
 A category that was included in your search has been deleted or deactivated.
 Your visibility settings have changed and you are no longer authorized to view some
data that was included in the search.
Your system administrator changed your public saved search. Once your public search is
changed by your system administrator, you are no longer the owner of that specific
search.
Saving a Search
Searches are saved per user. This means that only you and the system administrator
can edit and delete the searches that you created.
You can delete that search and create another one instead, with updated criteria.
To save a search:
1 Once you are satisfied with the search results you have received, click Save Search
in the Results page.
The Save Searches dialog box appears.
2 In the Enter a name for your saved search field, type a name for the search. The
name can consist of up to 30 alphanumeric characters, excluding the following
symbols: quotation marks ("), apostrophes ('), periods (.), semicolons (;), colons
(:), backslash (\), and arrow heads (< >).
3 To make the search available for all users, select the Public Saved Search
checkbox.
To make the search only available to you, do not select this checkbox.
4 Click OK.
The saved search is added to the My Searches folder in the Speech Analytics
application home page.
NOTE If you save a search more than once using the same name as the original saved
search, the original search parameters will be automatically overwritten.
To rename a saved search see “Renaming a Saved Search” on page 124.
Renaming a Saved Search

1 In the Speech Analytics application home page, click Manage My Searches.
The Manage My Searches window appears.
2 Click the Rename icon ( ) associated with the search you want to rename.
The Rename Saved Search window appears.
3 Enter the new name and click OK.
4 Click Close to return to the home page.

Chapter 4 - Contact Search and Analysis Using your Search History
Changing a Private Saved Search to a Public Saved Search

In order for a search to be available to all users it must be a Public Saved Search.
To change a Private Saved Search to a Public Saved Search:

The Manage My Searches window appears.
2 Click the Private Saved Search icon ( ) associated with the search that you
would like to make public.
Once you click the Private Saved Search icon, it turns the search into a Public Saved
Search, and the icon changes into the Public Saved Search icon ( ).
To toggle between the Public Saved Search and the Private Saved Search, click the
Public Saved Search icon.
NOTE The user who created the Public Saved Search or a user with Speech Analytics and
Assignment Manager permissions are the only users with the permissions to
change a Public Saved Search into a Private Saved Search.
Deleting a Saved Search

2 Click the Delete icon associated with the Saved Search you want to delete.
The Confirm dialog box appears with the following question:
Are you sure you want to delete this saved search?
3 Click Yes to confirm.
4 Click Close to return to the home page.
Using your Search History

The Speech Analytics application keeps track of your activities during the search and
shows them on the History bar of the Result page. Each entry on the bar displays your
previous search criteria and the number of found results. This information clearly
represents your search flow and helps you better understand the results displayed. For
more information, see “History Bar Items” on page 126.
The following is an example of the History bar.

Chapter 4 - Contact Search and Analysis Using your Search History
As you can see in the example, the following search activities took place:
1 First all contact segments (a total of 10516) were retrieved by clicking the Find
button in the home page, without any keywords entered.
2 After that, a search for the “thank you” phrase within these results was performed
and brought 1550 contact segments.
3 The last search operation was filtering, which reduced the number of results to 65.
Each entry on the History bar also serves as a link that allows you to return to previous
results in your search.
To return to the previous search results:

 On the History bar of the Result page, click the required entry. Once you clicked a
previous search item, the History bar does not longer display subsequent searches.
The search history is cleared as soon as you return to the Speech Analytics home page
or start a new search by clicking the New Find button.
History Bar Items

The following table describes all the items that can appear on the History bar.
Item Description Comment
Keyword(s) The keywords you entered as your

search criteria.
Category The category you selected from the

home page or from a chart. When you
place your mouse over the name of
the category, a tooltip appears
showing any filters that were set as
part of the category's definition. It
does not show any filters you set.
Filters Set This item appears when you set filters Move your mouse over the Filters
from the Results page. Changed text to view the filters you
set.
Filters Cleared This item appears when you reset

filters from the Results page.

Chapter 4 - Contact Search and Analysis Using Charts
Item Description Comment
All contacts This item appears when you set filters Clicking on this item will display a list
from the home page (followed by of all available contacts.
Filters Changed), and when you click
Find in the home page without first
entering keywords.
Saved search name This item appears when you select a To see any filters that were set as part
saved search. of the saved search, move your
mouse over Filter icon above the
results list.
Selected filter, (for The selected filter appears when a

example, Date 4 April 08) chart column has been selected.
Using Charts
Charts show statistical information about all transcribed contacts or the specific subset
of contacts you have retrieved from the Results page. The charts help you identify
trends in your contacts, for example, that a certain problem seems to happen during the
evening shift, or that words indicating dissatisfaction are highly correlated with your
technical support center.
Two bar charts covering a range of contact-related and category-related data are
displayed at the bottom of the Speech Analytics home page and in the Results page. By
default, the chart on the left displays the Date chart while the chart on the right is the
Category chart.
You can use the Show Legend icon to show each column's title in a legend at the
side of the chart rather than beneath the chart.
This topic discusses:
 Viewing the Number of Contact Segments in Charts, page 127
 Viewing Percentage Values in Charts, page 128
 Drilling-down into Charts, page 129
 Chart Descriptions, page 130
If a chart does not appear, refer to “Troubleshooting” on page 229.
Viewing the Number of Contact Segments in Charts

Before a search or filtering has been performed, the values displayed in the charts
represent the total number of transcribed contact segments. Once a search query has
been performed, the chart shows the number of segments that match the query.
For example, before a query has been performed, in the Date chart you may see that
there are a total 20 transcribed contacts in the Speech Analytics application from the 7th
of June. If you then perform a query for the word “login” and then look at the Date chart
in the Results page, you may see that 10 results were found from the 7th of June. This

means that out of the 20 contact segments recorded on the 7th of June, 10 of the
segments contained the word “login”.
In any chart, you can view the number of contact segments in each bar.
To view the number of contact segments in each bar:

 Hover your mouse over the bar in the chart. The number of segments is displayed in
a tooltip.
The following is a tooltip example on the Duration chart in the home page, without
any queries performed. The tooltip shows that the total number of short contact
segments is 147.
Viewing Percentage Values in Charts

When viewing results in the Results page after performing a query, you can toggle the
chart display between numeric or percentage values. Both values are calculated and
displayed per each bar in the chart. While the numeric value represents the number of
segments in the bar, the percentage value shows the percent of these segments out of
the total number of segments available for that bar.
The percentage is calculated by dividing the numeric value of the bar displayed in the
charts after performing a search with the total number of contacts available for the bar,
irrespective of the query that was performed. For example, if there are a total of 20
contacts in the Speech Analytics application from the 7th of June, 10 of which match the
search, the percentage displayed in the date chart for the 7th of June will be 50% (10/
20). For example, the value 10 in the Date chart represents 50% of all contacts
recorded on the 7th of June.
For more information about charts, see:
 Drilling-down into Charts, page 129
 Chart Descriptions, page 130

NOTE When you select a filter without entering keywords, the percentage value is not
calculated as the value in the results charts already represents the total number of
contacts that match the filter, and thus the percentage will always be 100%. For
example, if there are 20 contacts from the 7th of June and you drill down on the
7th of June in the chart (or set it as a filter and click Find), the resulting chart will
also display 20 contacts.
If percentage values for a chart are available, the Show percentage of total icon is
displayed to the left of the chart name.
To display the chart values as a percentage of total:

1 Click the Show percentage of total icon . The chart reloads to display the
percentage values for each bar.
2 Hover your mouse over the bar to see the percentage of segment out of the total
number of segments.
The following is a tooltip example on the Duration chart in the Results page, after a
query was performed. The tooltip shows that short segments constitute 23.8% of
the total number of short contact segments (35 out of 147).
Drilling-down into Charts

You can click on any column in a chart to view a set of contacts that match the value
that you selected. If you select a group that contains subgroups (rather than agents),
the subgroups will be displayed. Clicking a subgroup will display its subgroups, until the
group you click contains agents, and then the agents will be displayed in the chart.

The Speech Analytics application behaves differently depending on the chart you select:
 Clicking a bar in a chart (excluding the Category Root Cause, Category,
Category by impact ($) and Keyword charts) sets the bar's value as a filter. In
this case, when performing a new keyword search from the Results page (by
clicking Find New), the previously selected chart column will be used as a filter
when performing the search. To remove the filter, select Reset Filters. For more
information on filtering, see “Defining Search Filters” on page 93.
 Clicking on a column in the Category Root Cause, Category, Category by impact ($)
and Keyword charts does not affect the filters that have been set. In this case, when
performing a new keyword search from the Results page (by clicking Find New),
the previously selected chart column (for example, category or keyword) will be
ignored when performing the search.
Chart Descriptions
You can use the drop-down list that appears above each chart to choose from the
selection of the following available charts:
NOTE Each chart contains a horizontal line that represents the average number of
contacts/segments for the selected chart.
Category Root Cause

This chart displays the relevance of the categories to the search results. This chart is not
available from the Speech Analytics home page and is only available from the Results
page when a keyword or category search has been performed (i.e. if you perform only a
filter search this chart is not available). The relevance of each category is ranked on a
scale from zero to one, with the highest rank assigned to the most relevant category.
When a category is used to search for contacts, the relevance of the category you
selected will always be 1.0, as you are viewing the entire contents of the category.

However, rather than displaying the chart on a scale of zero to one, Speech Analytics
“zooms in” on the chart so that you can focus on the relevance of the other categories.
When you place your mouse over the category you selected for the search, the tooltip
will show you that the actual value is indeed 1.0.
NOTE If you have been assigned either the Speech Analytics According to Group right or
the Speech Analytics Restricted right in the Assignment Manager, the category you
selected may not be ranked as one as you may only have access to part of the
category's contacts.
Category
This chart displays the number of contacts associated with the categories defined in the
system. When a search has been performed and/or filters have been set, the number of
contacts associated with each category is based on the search results, or the set of
contacts that match the filter.
Category by impact ($)

This chart displays categories according to their potential dollar impact. The category's
per contact value is part of the category's definition, and the total value is calculated by
multiplying each contact's ranking score with the defined value, and adding up the
results.
So, if you have a category called Customer Complaints that you have assigned the value
of $4, and Speech Analytics finds 100 contacts that are associated with this category,
the category's $ impact could potentially reach $400 (depending on the rank of the
contacts).

Keyword
When performing a keyword search, this chart displays the number of contacts found for
each of the keywords. When using categories to search for contacts, this chart displays
the number of contacts found for the keywords that are defined in the category you
selected.
This chart is only available from the Results page, and only appears if a keyword/
category search has been performed (for example, if you only perform a filter search or
search for all contacts in Speech Analytics, this chart is not available).
/ category
Date
In the home page, when no filters have been set, this chart displays the total number of
available transcribed contacts, per date.

When a filter has been set or a search has been performed, the chart displays the
number of contacts for the search results or defined filters.
When the chart needs to display a range of up to 14 days, each day is displayed. For a
range of between 15 and 92 days, the dates are displayed as weeks, and for ranges
above 92 days, the dates are displayed as months.
Time of Day
available transcribed contacts, per time of day the contacts were recorded.
The contacts are grouped into the following times of day:
 03:00-06:00
 06:00-09:00
 09:00-12:00
 12:00-18:00
 18:00-03:00
These definitions are defined in the Enterprise Manager.
The contacts are displayed according to agent's local time at the time of recording.

Agent
available transcribed contacts, per agent.
The chart displays up to seven agents, starting with the agent that had the most
contacts. To see the next batch of agents, click the Show next results icon . To jump
to the agents with the lowest number of contacts, click the Show last results icon .
Agent Group
available transcribed contacts, per group.

The chart displays up to seven groups, starting with the groups that had the most
contacts. To see the next batch of groups, click the Show next results icon . To jump
to the groups with the lowest number of contacts, click the Show last results icon .
Duration
available transcribed contacts, per contact duration, displayed in the following default
ranges: Very short (0-35 seconds by default), Short (35 secs. -1 min. 55 by default),
Average length (1:55-3:00 by default), Long (3:00 - 5:00 by default), Very long (5:00 -
90:00 by default). The definitions for these ranges are defined and can be changed in
the Enterprise Manager where the instance is created.
Above the chart, you can see the average duration of contacts, in seconds, to which
segments included in the chart belong.
Direction
available transcribed contacts, per direction.

ANI
available transcribed contacts, per ANI.
The chart displays up to seven ANI values, starting with the ANI values that had the
most contacts. To see the next batch of ANI values, click the Show next results icon
. To jump to the ANI values with the lowest number of contacts, click the Show last
results icon .
DNIS
available transcribed contacts, per DNIS.

The chart displays up to seven DNIS values, starting with the DNIS values that had the
most contacts. To see the next batch of DNIS values, click the Show next results icon
. To jump to the DNIS values with the lowest number of contacts, click the Show last
results icon .
PBX ID
available transcribed contacts, per PBX ID.
The chart displays up to seven PBX ID values, starting with the PBX ID values that had
the most contacts. To see the next batch of PBX ID values, click the Show next results
icon . To jump to the PBX ID values with the lowest number of contacts, click the
Show last results icon .

Number of Conferences
available transcribed contact segments, per number of conferences that can occur in
contacts.
number of contact segments for the search results or defined filters.
Number of Holds
available transcribed contact segments, per number of holds that occurred in the
contacts.

Above the chart, you can see the average number of holds for contacts whose segments
are included in the chart.
Number of Transfers
available transcribed segments, per number of transfers that can occur in contacts.
Above the chart, you can see the average number of transfers for contacts whose
segments are included in the chart.

Total Hold Time

available transcribed contact segments, per total hold time, displayed in the following
default ranges: Very short (0-5 seconds), Short (5-10 seconds), Average length (10-20
seconds), Long (20 - 60 seconds), Very Long (above 60 seconds.).
Above the chart, you can see the average number of total hold time for contacts whose
segments are included in the chart.
Wrap-up Time
available transcribed contacts, per wrap-up time (for example, the additional time the
agent spent closing the contact after s/he finished talking to the customer), displayed in
the following default ranges: Very short (0-1 min), Short (1-2 mins.), Average length
(2-3 mins.), Long (3 - 4 mins.), Very Long (above 4 mins.).

Extension
available transcribed contacts, per extension.
The chart displays up to seven extensions, starting with the extensions that had the
most contacts. To see the next batch of extensions, click the Show next results icon
. To jump to the extensions with the lowest number of contacts, click the Show last
results icon .
Exception Contacts
In the home page, when no filters have been set, this chart displays the number of
contacts that met an “Exception” rule defined in the Rule Editor, for the total number of
available transcribed contacts. For example, an Exception rule can be created for all
contacts that contain more than four transfers. Contacts with more than four transfers

will be designated as “exception” contacts, will be pushed into the Exception Contacts
folder in the Customer Xperience application, and are included in the Is Exception chart.
number of exception contacts for the search results or defined filters.
Agent Talk Time Percentage

The Agent Talk Time Percentage chart is displayed only if Speaker Separation has been
enabled in your system. In the home page, when no filters have been set, this chart
displays the total number of available transcribed segments, per agent talk time
percentage.
The X-axis displays bars for the following default ranges: Very short (0-20%), Short
(20-40%), Average length (40-60%), Long (60-80%) and Very Long (80-100%).
The Y axis shows the number of segments in each group.
When a filter has been set or a search has been performed, the chart refers to retrieved
contact segments, and not to the entire volume of transcribed segments.
The agent talk time percentage can be calculated only for segments that were recorded
with Speaker Separation. Segments recorded without Speaker Separation, if such are
included in the result set, are omitted from the chart. As a result, the number of
segments in such a chart is lower than the number of the segments in the result set.
Above the chart, you can see the average agent talk time percentage calculated for
segments included in the chart. If segments recorded without Speaker Separation were
omitted from the chart (and therefore, also from the calculation), an asterisk appears
next to the calculated average.
TIP Hover your mouse over the asterisk to view a tooltip with an explanation on this
value's calculation.

You can also view the average agent talk time percentage in the query result metrics.
For more information about the query result metrics, see “Viewing Query Results
Metrics” on page 113.
Customer Talk Time Percentage

The Customer Talk Time Percentage chart is displayed only if Speaker Separation has
been enabled in your system. In the home page, when no filters have been set, this
chart displays the total number of available transcribed segments, per customer talk
time percentage.
The customer talk time percentage can be calculated only for segments that were
recorded with Speaker Separation. Segments recorded without Speaker Separation, if
such are included in the result set, are omitted from the chart. As a result, the number
of segments in such a chart is lower than the number of the segments in the result set.
Above the chart, you can see the average customer talk time percentage calculated for

You can also view the average customer talk time percentage in the query result
metrics, see “Viewing Query Results Metrics” on page 113.
Silence Time Percentage

available transcribed segments, per silence time percentage.
Above the chart, you can see the average silence time percentage calculated for
segments included in the chart.
When Speaker Separation is enabled, Silence time is detected and provided by the
recorder. When Speaker Separation is disabled, Silence time is provided by the Speech
Transcription process.

You can also view the average silence time percentage in the query result metrics. For
more information about the query result metrics, see “Viewing Query Results Metrics”
on page 113.
Talk-Over Time Percentage

The Talk-Over Time Percentage chart is displayed only if Speaker Separation has been
enabled in your system. In the home page, when no filters have been set, this chart
displays the total number of available transcribed segments, per talk-over time
percentage.
The talk-over time percentage can be calculated only for segments that were recorded
with Speaker Separation. Segments recorded without Speaker Separation, if such are
included in the result set, are omitted from the chart. As a result, the number of
segments in such a chart is lower than the number of the segments in the result set.
Above the chart, you can see the average talk-over time percentage calculated for

You can also view the average talk-over time percentage in the query result metrics, see
“Viewing Query Results Metrics” on page 113.
Custom Data and Conditional Custom Data

Up to 30 custom data and conditional custom data charts may appear in the list,
depending on how your system has been configured. The names of these charts are also
configurable and are defined when your system is set up.
The Speech Analytics application supports up to 25 contact custom data fields, per
contact, whose values come from CTI data obtained from the switch (for example,
Account Number, Transaction, Site, etc).
In addition, the Speech Analytics supports up to 5 conditional custom data fields, per
contact. These data fields contain predefined values that are based on customized rules
defined by the system administrator. For example, there may be a Customizable Custom
Data field called Line of Business. Your organization can define a rule whereby the Line
of Business that is displayed is dependent on the DNIS that is dialed by the customer.
The conditional custom data values to which you have access are assigned to you in the
Assignment Manager. So, your system may have five lines of business, but you may only
be assigned to three of the five. As a result, the chart that you will see will only show
these three values. However, if you were only assigned three of the five values, but you
were also assigned the right to “Search for any value”, you will be able to see all five
lines of business.

The chart displays up to seven Custom Data values, starting with the Custom Data
values that had the most contacts. To see the next batch of Custom Data values, click
the Show next results icon . To jump to the Custom Data values with the lowest
number of contacts, click the Show last results icon .
Star Ranking
This chart enables you to display the current result set of calls by their Star Ranking.
This chart is only available when creating a chart report. It is not available from the
Results List page. For more information, see “Creating Chart Reports” on page 209.
Uncategorized Contacts
This chart includes all of the contacts that do not meet any of the existing category
definitions.

Chapter 4 - Contact Search and Analysis Starting a New Search
Starting a New Search

You can start a new search from the Speech Analytics home page or from the Results
page. If any filters were used in the previous search, you need to decide whether you
want to keep them, and if not, reset them first. Otherwise, the filters will be applied to
your new search, and the search results will be filtered accordingly.
1 If you wish to reset the filters that were used in the previous search, click the Reset
Filters button. If you do not reset the filters, the results of your new search will be
filtered by them.
TIP To see the currently set filters, place your mouse over the Filter icon above the
results list to view the filters in a tooltip.
2 Use any of the search methods:

 Searching by Keywords or Phrases, page 67
3 Click the Find button when searching from the home page, or click the New Find
when searching from the Results page. The Results page appears displaying a list of
contacts matching your search criteria.
NOTE If you wish to further refine your search, you can continue “Searching within
Found Results” on page 121 while “Refining Search Results through Different
Search Methods” on page 122.

Chapter 5
Contact Playback
This chapter outlines the information required for understanding how to playback
contacts, by covering the following topics:
 Playing Back Contacts in Speech Analytics, page 150
 Understanding the Speech Analytics Player, page 151
 Transcription Area, page 154
 Speech Analytics Player without Speaker Separation, page 156
 Viewing Screens, page 157
 Opening a Contact, page 158
Chapter 5 - Contact Playback Playing Back Contacts in Speech Analytics
Playing Back Contacts in Speech Analytics

You can playback contact segments that have been retrieved in the Results page or in
the category definition page. While a contact segment is being played back, you can
view its screens and simultaneously view a transcript of the call.
NOTE You can also open contact segments in the Portal. See “Opening a Contact” on
page 158 for more information.
To playback a contact segment:

1 In the Results page or in the Category Definition page, click the Play this
contact button for the segment you want to playback.
The Play this contact button turns into a Close the player button , and the
Speech Analytics Player appears at the bottom of the Results page, replacing the
charts.
For a detailed explanation of the player, see “Understanding the Speech Analytics
Player” on page 151.
NOTE If the Play this contact button does not appear, refer to “Troubleshooting” on
page 229.
If your system is set to playback audio over speakers or headphones, the audio is
immediately played back.
2 During audio playback, use the following methods to quickly move to different
portions of the audio:
 Click any term callout above or below the audio waveform to move to the term
location in the audio.
 Click a word in the transcription to jump to its location in the audio.
3 Upon listening to the audio, click the Close the player button .

Chapter 5 - Contact Playback Understanding the Speech Analytics Player
Understanding the Speech Analytics Player

The following is an example of the Speech Analytics player.
NOTE The appearance of the Speech Analytics Player depends on the way the segment
was recorded - with or without Speaker Separation. This picture illustrates the
Speech Analytics Player for a contact segment recorded with Speaker Separation.
For segment recorded without Speaker Separation, the player does not contain
any Speaker Separation information. See “Transcription Area” on page 154.
The Speech Analytics Player can display the following elements:

 Audio Waveform, page 151
 Callouts, page 152
 Speaker Separation Bars, page 153
 Progress Bar, page 153
 Playback Control Buttons, page 154
 Category Bar, page 154
 Transcription Area, page 154
Audio Waveform
A single wave in the player represents the audio graphically.
NOTE For stereo-recorded contact segments, the audio waveform combines the audio
recorded from both channels.
Parts of the waveform are color-coded to indicate the following characteristics of the
conversation:
For contact segments recorded with Speaker Separation:

 Blue color: the agent is speaking.

NOTE When there is an internal conference call where there are three parties in the call
(the customer and two agents), the customer is regarded as one side of the
conversation while all agents are combined on the other - agent's - side of the
conversations.
 Green color: the customer is speaking.

 Aqua color: either both the agent and the customer are speaking at the same
time, or the system could not identify the speaker.
 Gray color: silence
For segments recorded without Speaker Separation:

 The Blue color represents the voice of both the agent and the customer.
For all segments:

 The Yellow color represents emotional parts of the conversation, regardless of the
speaker.
TIP If you do not want to view the emotional parts of the contact in the audio
waveform, click the Hide Emotions icon .
 Voice energy levels are represented as a series of peaks; the absence of the peaks
indicates silence.
Callouts
Callouts point to the location of keywords that you searched for (or defined in the
category you selected) on the waveform. You can click a callout to jump to the relevant
place in the audio. The appearance and position of callouts is different for contact
segments recorded with Speaker Separation and for contact segments recorded without
it.
For segments recorded with Speaker Separation: callouts are positioned and color-
coded by the speaker:
 Callouts to the terms spoken by the agent are positioned above the waveform, in
blue (for example, tracking number).
 Callouts to the terms spoken by the customer are positioned below the waveform, in
green (for example, tracking number).
 Callouts to terms that were detected during talk-over time, when both the agent
and the customer were talking at the same time, appear twice - both above and
below the waveform - but pointing to the same location. The text color in such
callouts is aqua (for example, tracking number).
For segments recorded without Speaker Separation:

 All callouts appear above the audio waveform, in blue, regardless of the speaker.

For all contact segments:

 There is no emotion indication in callouts.
TIP You can know that the term was emotional if the callout is pointing to a yellow
section of the waveform. In the example above, the term "accidents", which was
spoken by the agent and therefore appears above the waveform in blue, was
emotional because its callout is pointing to a yellow area of the waveform.
 When a term callout is active - either because you selected it, or because the
playback has reached the location in the conversations where the term was spoken
- the colors of the callout change to the inverse ones. Once the player progresses
past the associated area of conversation, the callout is reverted to its default colors.
Speaker Separation Bars

Speaker Separation bars are displayed only for contacts recorded with Speaker
Separation.
Speaker Separation bars are two horizontal bars located below the waveform. The top
bar, in blue ( ), represents the agent, and the lower bar, in green ( ),
represents the customer. Each bar is filled with color when the respective participant is
talking. Talk-over time, when the agent and the customer were talking at the same time,
is marked in both bars with the aqua color ( ). Non-colored areas on the bars
correspond to silence in the audio waveform.
Speaker Separation bars, together with the audio waveform, give you a full picture of
the dialog between the agent and the customer. For example, the waveform does not
indicate the speaker during emotional portions of the conversation, but you can obtain
this information from the Speaker Separation bars.
Progress Bar
The slider on the progress bar advances as the playback of the segment progresses. You
can drag the slider to move to a different place in the conversation.
Above the progress bar, on the right, you can see the counter, which increments during
the audio playback, and the total duration of the contact segment.
If the contact segment which is being played back was encrypted, a lock icon is
displayed above the progress bar, on the right. If only some components of the contact
were encrypted (for example, audio is encrypted and screens are not), the lock icon is
half-transparent .

Playback Control Buttons

Use these button to control audio playback.
Buttons Description
Play/Pause Plays the audio after pressing Pause or pauses the audio
after pressing Play.
Mute Mutes the audio.
Volume Adjusts the volume.
View Screens Displays the agent's screens in a separate window. To

read more about viewing agent's screens, see “Viewing
Screens” on page 157. The button is disabled if the
segment does not contain screens.
Stop Stops the playback and rewinds the call back to the
beginning.
Previous Keyword Jumps the playback to the previous instance of a

keyword.
Next Keyword Jumps the playback to the next instance of a keyword.
Hide Emotions Removes the emotional color-coding (yellow) from the

audio waveform and the transcription area.
NOTE The adjustments you make to the volume, as well the display of the screen
window, are maintained for the playback of all the segments retrieved for your
current search. However, when you perform another search and then play
segments, the default volume settings are restored, and the screen window is
hidden.
Category Bar
The category bar shows the names of the categories this segment belongs to and its
rank in each of them (from three to five stars). This area is color-coded according to
category colors. For example, if a category's color is red, then the category's name and
the segment's ranking in the category appear in red ( ).
If you performed a search by category, that category is always displayed with 5 stars.
Yellow stars represent the Emotions category, where the number of stars represents
the level of emotion in the conversation (from three to five stars, where five stars
indicate the most emotional conversations).
Transcription Area
The transcription area of the Speech Analytics player displays the text of the
conversation between the agent and the customer. This area also serves as a visual map
of the conversation, indicating the location of category terms and keywords you

searched for. For segments recorded with Speaker Separation, the transcription area
also displays speaker indicators.
You can click any word in the transcription to jump to its location in the audio.
The following conventions are used in the transcription area:
 Speaker color-coding and indicators (only for segments recorded with Speaker
Separation):
- The up arrow indicates that the agent is beginning to speak.
- The down arrow indicates that the customer is beginning to speak. The text
spoken by the customer appears with light-green shading.
- The diamond symbol indicates the beginning of the call portion when both the
agent and the customer are speaking simultaneously. The text spoken by both
the agent and the customer appears with light-blue shading.
 Keywords or phrases specified in the search or defined in the selected category are
underlined.
 Words and phrases (including keywords that were searched for or included in the
specified category) are color-coded according to the category to which they belong.
In addition, each word is in one of three shades of the category's color to represent
the success rate of the transcription. The darker the color the higher the success of
the transcription.
For example, if the Customer Retention category color is red, each term belonging
to the category will be in one of three shades of red.
 Each word in the transcription area that does not belong to a category is displayed
in one of three colors - gray, dark gray, or black - to represent the success rate of
the transcription. The darker the color the higher the success of the transcription. If
a word could not be transcribed with enough confidence of success, it is replaced
with an asterisk.
 As the audio plays, each transcribed word is highlighted, in synchronization with the
audio.
For more details, see “Speaker Separation” on page 18.

Chapter 5 - Contact Playback Speech Analytics Player without Speaker Separation
Speech Analytics Player without Speaker

Separation
The appearance of the player depends on the availability of Speaker Separation data for
the segment being played.
NOTE For more details, see “Speaker Separation” on page 18.
The following is an example of the Speech Analytics Player for segments recorded
without Speaker Separation.
For segments recorded without Speaker Separation, the player does not display any
speaker indications:
 Audio Waveform
In the audio waveform, two colors represent voice - blue for non-emotional parts of
the conversation and yellow for emotional ones. Voice energy levels are represented
as a series of peaks: the absence of the peak indicates silence.
NOTE For stereo-recorded contact segments, the audio waveform combines the audio
recorded from both channels.
 Callouts
All callouts are displayed above the audio waveform, in blue, regardless of the
speaker.
 Speaker Separation Bars
Speaker Separation bars are not displayed for segments recorded without Speaker
Separation.
 Transcription Area
There are no speaker color-coding and indicators in the transcription area.

Chapter 5 - Contact Playback Viewing Screens
NOTE The functionality of the Speech Analytics Player remains the same regardless of
the way the contact segment was recorded - with or without Speaker Separation.
For a full description of the Speech Analytics Player and its full functionality, see
“Understanding the Speech Analytics Player” on page 151.
Viewing Screens
When you click on the View Screens icon in the playback area, the screens
associated with this call are displayed in a separate window.
NOTE If there are no screens associated with this call, the Screen Exists icon will not
appear in the Results list.
The Screen Exists icon will appear even if the screen file has already been
deleted from the system.
From the screen window, you can perform the following actions:
 Configuring the Screen Display, page 157
 Controlling Audio Playback while Viewing Screens, page 157
 Force the screen window to always appear on top of any other selected screen.
Configuring the Screen Display

By default, the agent's screen is reduced to fit into the window. To view the screen in its
original resolution, click the Actual Size button . The recorded screen appears in its
original resolution and the Actual Size button appears as follows . Clicking the
button again shrinks the screens to fit the window.
You can now resize the window to suit your viewing needs. For example, you can place
the screen window at the top of the screen and leave the play controls in the playback
area visible, or you can place the screen window at the bottom of the screen and
continue playing back different calls from the Results window without needing to move
the screens each time.
Each time you position and resize the screen window, the window remains the same size
and position for all subsequent contacts that you open from the Results window, until
the next time you resize/reposition or until you leave the Results page altogether (either
to return to the home page or to navigate to a different page).
Controlling Audio Playback while Viewing Screens

An alternative way of viewing the screens is clicking the Maximize button and
controlling the audio playback by right-clicking anywhere on the screen and using the
menu to play, pause, stop and adjust the volume of the playback.
If you want the screen window to remain on top of all other windows on your desktop,
even when you click on another window, click the Always on Top button .

Chapter 5 - Contact Playback Opening a Contact
To remove the Always on Top setting and let selected windows cover the screen window,
click the Always on Top button again .
Opening a Contact
Once you have performed a search in the Speech Analytics application, you can open
contact segments in the QM & Analytics Portal applications. These are the Quality
Monitoring, Contacts, or Customer Xperience applications.
Opening a contact in each specific application gives you access to its functionality. For
example, after opening a contact in Quality Monitoring, you can perform an evaluation,
assign a flag, or forward the contact to an agent.
If you have access to more than one application, you can select which of these
applications to use as default for playing back contacts from Speech Analytics. This can
be done via the Portal's preferences page which can be accessed by clicking the
Preferences button from the Portal home page.
 Click the start time in the Start Time and Data column of the contact you want to
playback. The selected contact opens in your assigned application.

Chapter 6
TellMeWhy Analysis
This chapter outlines the information required for understanding how to use root cause
analysis, by covering the following topic:
 Viewing TellMeWhy Analysis, page 160
Chapter 6 - TellMeWhy Analysis Viewing TellMeWhy Analysis
Viewing TellMeWhy Analysis

TellMeWhy™analysis provides root cause analysis that helps identify underlying drivers
of performance and determine opportunities for improving efficiency and quality in your
business.
The information Speech Analytics holds about the actual content of the call can provide
insights into the issues discussed in your calls. The Speech Analytics application can
thus suggest the root cause driving the set of contacts you retrieved, by identifying up
to 10 different clusters of root cause keywords and phrases that were deemed to be
most significant in these calls. Each cluster is also displayed as a percentage compared
with total number of contacts matching the query.
Each cluster can contain up to 20 phrases, keywords or NEAR phrases (for example,
credit NEAR card) by default. These terms can help you to identify the root cause of your
customers' most important issues, without needing to listen to each call. For example,
when analyzing the Customer Complaints category, you may discover that the cluster
with the highest percentage contained the words "bill" and "mistake", thus providing
you with immediate insight into the root cause of your customers' complaints.
The Speech Analytics application suggests the most relevant keywords or phrases in the
cluster of results you are currently viewing, and enables you to add them to your current
keyword search and receive a subset of results that actually reflect the underlying issues
in the calls. You can either select individual phrases or entire clusters. Playing back
these contacts will help you zero-in on the actual cause behind this business issue.
The TellMeWhy™ button is not available if you were not assigned the "Search for
keywords" right in the Assignment Manager, as the tool automatically enters selected
words into the Search field.
TIP You may consider creating new categories based on the TellMeWhy™ analysis.
To receive root cause suggestions:

1 Click the TellMeWhy™ button in the Results page to receive a list of up to 10
clusters of the most relevant terms that match the current list of contacts.
The following window is an example of the Root Cause suggestions that will appear.

The bottom of the window displays a chart showing the percentage of each cluster
compared with total number of contact segments matching the query. For example, if
your query returned 200 contacts, and Root Cause 1 comprises 50 contacts, the chart
will show the percentage as 25. To view the exact percentage value, hover your mouse
over the bar.
2 To view the Root Cause, place your mouse over the Root Cause title (for example,
Root Cause 1). The score that appears indicates the relevance of each Root Cause to
the set of contacts you retrieved (that is, the dataset).
This relevance score represents the ratio between the contacts in the Root Cause
group, and their relationship to the contacts in the entire search results. The higher

the value the more relevant the Root Cause is to the set of contacts you retrieved
from the search results.
A minimum relevance score is 2.5-3 and any number above 5 indicates a high
relevancy ratio.
NOTE When the dataset is too small, the relevance of the Root Cause analysis to the set
of contacts you retrieved is weak. As a result, there will not be enough statistical
contact information and a root cause cluster will not be displayed. When this
happens the following message appears
No root cause clusters have been identified for these search results.
3 Add the terms that you are interested in analyzing further to the Search field:
a. Select the terms by clicking on them. You can also select a cluster by clicking a
bar in the chart.
b. Click Add to Search. The selected terms are added to the Search field. You can
now add operators between the words if required, and either search within the
current search results or perform a new search. For more details, see “Searching
by Keywords or Phrases” on page 67.

Chapter 7
Category Definition
This chapter outlines the information required for understanding how to create and work
with Categories, Impact 360by covering the following topics:
 Speech Analytics Categories, page 164
 Category Builder, page 165
 Understanding the Category List, page 166
 Understanding the Category Definition Page, page 168
 Creating Speech Analytics Categories, page 170
 Defining Script Adherence and Non-Script Adherence Categories, page 172
 Selecting Category Colors, page 177
 Defining the $ Impact of a Category, page 178
 Category Terms, page 179
 Defining the Rank Threshold of a Category, page 195
 Defining Category Filters, page 195
 Defining Category States, page 198
 Number of Categories, page 199
 Copying Categories, page 201
 Editing Categories, page 202
 Synchronizing Categories, page 203
 Deleting Categories, page 204
 Exporting and Importing Categories, page 205
 Creating Category Definition Reports, page 207
Chapter 7 - Category Definition Speech Analytics Categories
Speech Analytics Categories

Speech Analytics categories are an efficient way of finding calls pertaining to a specific
business issue among the total scope of contacts, such as customer complaints, billing
issues, product feedback, or repeat calls. Once a category is created, any contact that
meets the category definition is automatically assigned to it. Depending on its properties
and terms used, a contact can be assigned to one, several or no categories.
Categorization can provide quantitative information (such as an increase in customer
complaints about late shipments), as well as qualitative information (such as a set of
calls in which customers state that they are switching vendors because of a better
pricing offer).
Speech Analytics also includes a predefined, yet customizable, Emotions category that
helps identify calls that may be more emotional than average. For more information, see
“Searching for Contacts that Indicate Emotion and Dissatisfaction” on page 106.
In addition, Speech Analytics Categories include a collection of Uncategorized
contacts, that includes all of the contacts that do not meet any of the existing category
definitions. Uncategorized contacts allows you to search for and analyze (using existing
contact details), contacts in the Speech Analytics index that are not assigned to any
category.
Category Builder is a dedicated tool for maintaining category definitions that
incorporates several automated methods that help define effective categories more
quickly and efficiently. For information, see “Category Builder” on page 165.
Categories can be shared with other integrated applications, thus enriching original sets
of data available in these applications. For instance, categories are used as search
criteria in Quality Monitoring and serve as filters in the Reports application. By defining
the category's states (that is, Published, Active, or Inactive), you can control which
categories are available in Speech Analytics only, in other applications, or are not in use
at the moment.
Category Trends and category trend analysis show how the performance of a category
changes over time. For more details, see “Category Trends” on page 25.

Chapter 7 - Category Definition Category Builder
Category Builder
Category Builder is a dedicated tool for maintaining category definitions. After you
provide a basic key term, the Category Builder can suggest which terms and phrases
should be used in the specific environment, helping to continuously refine categories
and pull together additional contacts of relevance. These suggested terms and phrases
help create a category definition that reflects how your agents and customers actually
express themselves in specific interactions, in many cases using terms and phrases that
you would not otherwise know.
The functionality of the Category Builder can reduce the time required for creating
effective business categories from weeks or months to a few days or hours.
There are two main pages in the Category Builder:
 Category List
In the Category list, you can view a detailed list of all existing categories. From this
page, you can create new categories, access definitions of existing categories for
editing, or delete categories. You can also export and import category definitions or
create reports about them. See “Understanding the Category List” on page 166.
 Category Definition Page
In the Category Definition page, you can create and edit categories. See
“Understanding the Category Definition Page” on page 168.
NOTE If an alert was defined to be sent when a category exceeds a defined threshold, an
alert message (email or pop-up) will be sent to all defined users. For information
on alerts see “Speech Analytics Alerts” on page 21.
Opening the Category Builder

You can open the Category Builder from the Speech Analytics toolbar displayed at the
top of the application window.
The Category Builder button is available only when you are working in the Trend
Analysis tab or in the Home Page. If you are working in the Results page, this button
is not available.
To open the Category Builder:

1 Verify that you are working in either the Trend Analysis tab or the Home Page.
2 Click the Category Builder button.
A list of categories is displayed.
For more information, see “Understanding the Category List” on page 166.
To return to the Category Builder from the Category Definition page:

1 Click the Category Builder link in the top left corner of the window.
2 Click the Back to Category List button in the top right corner of the page.

Chapter 7 - Category Definition Understanding the Category List
Understanding the Category List

When you open the Category Builder, a detailed list of all existing categories in the
system is displayed.
The following is an example of the category list.
Column Value Displayed

Header
This column displays an icon showing the color chosen for the particular
category.
Check box Select the check box to mark the categories that you want to:
 Export.
For details, see “Exporting and Importing Categories” on page 205.
 Include in the category definition report.
For details, see “Creating Category Definition Reports” on page 207.
Category The name of the category.

The category name serves as a link to its definition.
Click the category name to view and edit its definition in the category
definition page.
For details, see “Understanding the Category Definition Page” on
page 168.
Last Update The last date the category definition was updated.
Speech Analytics supports the US date format only (MM/DD/YYYY).
Updated By The user who last updated the category.

Chapter 7 - Category Definition Understanding the Category List
Column Value Displayed

Header
Terms Various keywords, or terms, that make up the selected category.

The terms in the category are marked as follows:
 Terms that are defined as Very Important are preceded by a plus
sign (+).
 Terms that are defined as Avoid terms are preceded by a minus sign
(-).
 Phrases are surrounded by double quotes, inside parentheses.
 If a category includes filters, or contains multiple keywords whose list
is too long to fit one line, the cell appears collapsed, with a plus sign
(+) at the beginning and an ellipsis (...) at the end.
For details on viewing the entire list, see “Viewing Collapsed Details” on
page 112.
State The appearance of each category in the Category Builder is determined by

its state. A category can have one of the following states:
 Published categories appears in bold black font (PUBLISHED).
 Active categories appears in regular black font (ACTIVE).
 Inactive categories appears in regular grey font (INACTIVE).
The state of a category determines whether the category data is
synchronized with system databases. When you modify a Published
category or when you modify the fields of a published or active category
that is part of a Published category's filter, you can decide whether to
synchronize the category with the databases. When you delete a category
synchronization is automatic.
A category can have one of the following Synchronization states:
 indicates that the category is in the process of synchronization.
 indicates that synchronization failed.
 indicates a problem with the connection to the database.
 indicates that the synchronization status is unknown due to an

upgrade from a previous version. That is, category results from a
previous version are not automatically synchrnonized in with the
results from the current version.
For more information on category states, see “Defining Category States”
on page 198.
Click the Edit icon to view and edit the category's definition in the
category definition page. For details, refer to Editing Categories.
Click the Delete icon to delete the selected category definition. For more
details, see “Deleting Categories” on page 204.
This icon is not available for the Emotions category, which cannot be
deleted.

Chapter 7 - Category Definition Understanding the Category Definition Page
What do you want to do?

 Creating Speech Analytics Categories, page 170
 Editing Categories, page 202
 Deleting Categories, page 204
Understanding the Category Definition Page

When you click the New Category button, a category's name or the Edit icon in the
category list, the category definition page is displayed, with details of the category.
From this page, you can view the category definition and change it if required.
For detailed information, see the “Category Definition Page Legend” on page 168.
Category Definition Page Legend

 The top bar displays the category’s general properties, such as its name, color, $
Impact value, and rank threshold and allows you to change them. This area also
enables you to view and define the category’s state. For more information, see
“Creating Speech Analytics Categories” on page 170.
 The Filter area indicates whether any filters have been set for the category. Hover
with the mouse over this area to preview the filter definition in a tooltip. To view a
complete filter definition or change it, click the Define Filter button in the Classic
View tab. For more information on defining filters, see “Defining Category Filters” on
page 195.
 The My Term area enables you to introduce new terms for your category.
Autocompletion is automatically activated as you type in words.
 Click the Suggest For Term button to see the following:
 In the lower area of the window - a list of retrieved contact segments that
contain the term.
 In the Suggested Term area - a list of context-based suggestions for the term
appearing in the field in the My Term area.
 If you are working in the Graphical View, the suggested terms are also displayed
in the tree.
Using the Suggest For Term button is part of your category term definition both in
the Classic View and in the Graphical View. See “Category Term Definition: Classic
vs. Graphical View” on page 181.
 The Suggested Terms area displays a list of suggestions that the Speech Analytics
application offers either for your term, or for the entire category. You can add any of
the suggestions to your category.
 When the category definition page is opened for the first time, this area is empty. As
you start working on a category's terms, this area displays a list of contact

Chapter 7 - Category Definition Understanding the Category Definition Page
segments according to the operation that you performed. For example, if you click
the Suggest for Category button, this area displays a list of contact segments
meeting the category's current definition. Above the list, you can see its title (for
example, Contacts that belong to the Repeat Calls category) and the number of
found results.
You can playback any of the retrieved contact segments by clicking the Play this
contact button. For more details on playback, see “Understanding the Speech
Analytics Player” on page 151.
When you playback a contact segment in the Category Builder, it contains the
following three buttons:
 Back to Contacts - click this button to return to the contact segment list in the
Category Builder.
 Previous Contact - click this button to play the previous contact segment in the
list.
 Next Contact - click this button to play the next contact segment in the list.
 The list of contact segments that appears in the bottom of the page after you click
Suggest for Terms or Suggest for Category, can be filtered according to a specific
rank. For more information, see “Category Definition Page - Filter by Rank Level” on
page 169.
 Use the Classic View tab or the Graphical View tab to add terms to your category
and assign them the required significance level. By default, the Classic View tab is
displayed. For more information on each tab, See “Category Term Definition: Classic
The Classic View tab also enables you to set the filters for that category. See
“Defining Category Filters” on page 195.
Category Definition Page - Filter by Rank Level

Ranking is the process by which a contact segment is given an overall score in a search.
For more information, see “Rank” on page 109.
From the Rank column in the Results List, you can drill down and filter the displayed
result set by selecting a specific rank. Based on your selection, the Results List view is
automatically refreshed and only contacts that match your rank selection will be
displayed.
To filter the Results List according to a specific rank:

1 In the Results List view, click the Rank filter icon ( ).
The following Ranking menu appears:

Chapter 7 - Category Definition Creating Speech Analytics Categories
The menu includes the number and percentage of contacts in the search results that
match the corresponding rank.
2 Select the rank for which you would like to see results.
The Results List is refreshed and a list of contacts with the selected rank appears.
NOTE Even when a rank selection is made in the Rank menu and the Results List is
refreshed to reflect that selection, the Rank menu continues to reflect the
distribution of the entire Results List set.
When you perform another filter request (for example, Find, Find Within, Set Filters,
etc.), the Results List is refreshed with your new search request. The Rank column
contains the entire set of contacts relevant to your new search (that is, All Contacts).
Creating Speech Analytics Categories

You can only define Speech Analytics application categories if you have been assigned
the Manage Categories right in the Assignment Manager.
The total number of categories in one Speech Analytics instance depends on the type of
Speech Analytics Application Server you are working with. For additional information,
refer to “Number of Categories” on page 199.
NOTE To create a Script Adherence or a Non-Script Adherence category, refer to

“Defining Script Adherence and Non-Script Adherence Categories” on page 172.
To create a Speech Analytics Category, perform the following:

3 “Opening the Category Builder” on page 165.
4 Click the New Category button.
The category definition page appears.
5 In Category Name, enter a unique name for the category.
6 “Selecting Category Colors” on page 177.
7 “Defining the $ Impact of a Category” on page 178.
8 Define “Category Terms” on page 179.
9 “Defining the Rank Threshold of a Category” on page 195.

Chapter 7 - Category Definition Creating Speech Analytics Categories
10 “Defining Category States” on page 198.
NOTE When defining a category, you can return to its previous (last saved) definition. To
return a category to its previous definition, click Reset Category.
11 Click Save Category.

The Save Category button is enabled only when you make a change to the
category.
NOTE If the Date Range > From the Last filter value is required as part of the category
definition, you must change the From the Last value before saving the category.
For more information, refer to “From the last” on page 97.
12 Click the Back to Category List button to return to the list of categories.
13 Click the Exit Category Definition button to return to the Speech Analytics home
page.
After you created a new category (active or published), the following changes occur in
the application:
 All contact segments available in the Speech Analytics online database at the
moment are scanned, and the relevant ones are retroactively assigned to the
category. The number of contact segments for the new category is displayed by the
category name in the Contact Analysis tab. For more details, see “Contact Analysis
Tab” on page 64.
 You can search for contact segments according to the category. For more details,
see “Using Categories to Find Contacts” on page 104.
 The category's Index data trend is available. The trend starts from the earliest date
in the Speech Analytics online database. For more details, see “Adding Index Data
Category Trends in the Plot” on page 34.
 The category's historical trend is available. The historical trend starts from the date
of the first new incoming contact segments associated with this category. For more
details, see “Adding Historical Category Trends to the Plot” on page 33.
 In the Trend Analysis tab, in the category trend analysis results, the category will
appear at the end of the list. No trend analysis for the category will be available
until the minimal required data is accumulated. For details on minimal data
thresholds, see “Changing the Time Frame in the Plot” on page 40.

Chapter 7 - Category Definition Defining Script Adherence and Non-Script Adherence Categories
Defining Script Adherence and Non-Script

Adherence Categories
The Speech Analytics Script Adherence feature enables customers to measure an
agent’s adherence or non-adherence to a defined script. A script is a structured phrase
that dictates the way agents must express themselves. For example, proper call
opening, verification of customer information, offering new services and so on.
By evaluating calls for script adherence, customers are now able to:
 Ensure that agents meet external regulatory scrutiny and internal quality
requirements by highlighting any deviation from the required script.
 Improve agents quality and effectiveness.
 Identify agents with high/low adherence %.
By leveraging Speech Analytics categories to perform Script Adherence, the user can
identify calls that contain or do not contain a specific script. A category that is used to
retrieve contacts that contain a given script are referred to as Script Adherence
Categories.
A category for which the Inverse checkbox is selected can be used to retrieve contact
that do not contain the give script. Such a category is referred to as a Script Non-
Adherence Category or Inverse Category.
The following sections focus on the unique aspects of defining Adherence and Non-
Adherence categories:
 Defining a Script Adherence Category, page 172
 Defining a Script Non-Adherence Category, page 175
 Naming Conventions, page 176
 Retrieving the Complement Set of Calls, page 176
 Use Case Example, page 176
Defining a Script Adherence Category

The purpose of a Script Adherence category is to retrieve the calls that are likely to
contain a given script.
The Script Adherence category is characterized by numerous NEAR and phrase
expressions that are unique to the script. Generally, all expressions should be added to
the Important Terms list.
To define a Script Adherence Category:
1 Open Speech Analytics.
2 Click Category Builder > New Category.
See “Naming Conventions” on page 176.

3 Click Define Filter and set a filter (based on meta-data or another category),
targeting a call subset where a specific script should be included.
For more information about defining filters, refer to “Defining Category Filters” on
page 195.
4 Determine the script you want to search for.
For example: The privacy of this information is your right and our duty under federal
law and your decision will not affect your service".
5 Identify the key words and phrases of the script.
To simplify the process, underline the key words and phrases and strike out the
short conjunction terms.
For example: The privacy of this information is your right and our duty under federal
law and your decision will not affect your service.
NOTE A script that is too short or does not contain enough distinctive phrases is likely to
result in categories that contain only part of the calls that are expected to return.
6 Once the key elements of the script have been identified, prepare a series of NEAR
and phrase expressions that represent a sequential breakdown of those key
elements of the script.
We propose the following best practices for how to break down a given script:
 In order to identify the sequential flow of the script, use overlapping words and
phrases in each two consecutive expressions. For example, the NEAR
expression:
privacy NEAR information, will be followed by: information NEAR "your right",
and so forth.
 Make sure the list of expressions covers, as much as possible, all parts of the
script.

 Use NEAR expressions whose terms are separated by up to two words in the
original script.
For example, the following collection of expressions can be used to define the
current script:
privacy NEAR information
information NEAR "your right"
"your right" NEAR "our duty"
"our duty under"
duty NEAR federal
duty NEAR law
"federal law"
law NEAR decision
decision NEAR "not affect"
"not affect" NEAR service
7 Add all expressions from the previous step to the Important Terms section in your
category.
The goal of the category is to retrieve most of the calls that contain the script, while
at the same time, avoid as much as possible calls that contain only part of the
expressions but do not contain the script.
8 Test the precision of the category, meaning that the category does not include calls
in the higher rankings that do not contain the script.
To do so, invoke the Suggest for Category option and browse to the last pages of
the search results. Verify that those calls contain the script by either listening to the
calls or by verifying that the Keywords Found column contains most of the
expressions you defined in the category.
If you find higher ranking calls that do not contain the script, identify the
expressions that bring the non-relevant calls and refine them to be more specific to
the script.
For example: if the phrase "federal law" is used in contexts other than the above
script, you may use a more specific expression instead, such as:
"under federal law"
duty NEAR "federal law" NEAR your
TIP If the number of calls in the search results is too large to browse to the end, you
may perform this test on a sample of the calls by defining a stricter filter such as
date bounding.
9 Test the recall of the category, meaning that all of the calls that contain the script
are included in the category.
To do so, see “Defining a Script Non-Adherence Category” on page 175 and
“Retrieving the Complement Set of Calls” on page 176, to invert the category to see
the complementary set of calls for your category.

a. Invoke the Suggest for Category option on the inverse category, and browse
to the last pages of the search results.
b. Verify that those calls do not contain the script, by listening to the calls and
locating the keywords of the category in the transcription.
If the Inverse Category analysis includes higher ranking calls that contain the script,
add more expressions to the original category in order to capture these calls, as
described in step 5.
Listening to some of these calls may help to identify variations of the expressions
that the agents use or possible false transcription recognitions, for example: "you're
right" instead of "your right". Invoking Suggest for Category on the original
category and viewing the Suggested Terms list may also yield such expressions.
TIP If the number of calls in the search results is too large to browse to the end, you
may perform this test on a sample of the calls by defining a stricter filter such as
date bounding.
10 Repeat steps 7 and 8 until the category is stable.
NOTE When defining a category, the recommended default value for the View contacts
ranking from: option is 2. A call that has more stars in the result set is more
likely to contain the script. You may use a higher ranking value for the above
option to increase the precision of the results and thus retrieve only calls that are
more likely to contain the script.
Defining a Script Non-Adherence Category

The purpose of a Script Non-Adherence category is to retrieve calls that do not contain a
specific script.
The Script Non-Adherence category is characterized by a filter that specifies a subset of
calls that should contain the script and numerous NEAR and phrase expressions that are
unique to the script. Generally, all expressions should be added to the Important Terms
list. No expressions should appear in the Very Important Terms or Avoid Terms lists.
In general, for every category, the Suggest for Category button generates both the
search results of the category and a list of Suggested Terms that have high correlation
with the category's result set. Particularly, the suggested terms for an Inverse Category
are those terms that have high correlation with the inverse result set. Therefore, if this
feature is used on a Script Non-Adherence Category, the Suggested Terms list is
expected to contain terms that are dominant exactly in those calls that do not contain
the script, if there are such terms.
To define a Script Non-Adherence Category:

1 Define a Script Adherence category as explained in the “Defining a Script Adherence

Category” on page 172.
2 Select the Inverse Category checkbox.
Perform ing this step informs the system to retrieve the calls where the script
defined in step #1 was not identified.
3 Select a value for the View contacts ranking from: option.
A recommended value is 4. A call that has more stars in the result set is more likely
to not contain the script.
Naming Conventions
To differentiate between Script Adherence categories and other categories in the
different application screens (Contact Analysis Home page, Trends, Charts, Category
Builder), the use of a naming convention for Script Adherence categories is
recommended. For example, for a Script Adherence category add the prefix "Script" (for
instance, Script-verify caller) and for a Script Non-Adherence category add the prefix
"Inverse" (for instance, Inverse-verify caller) to the category name.
Retrieving the Complement Set of Calls

Assume you use the same query and filter for a Script Adherence category and for a
Script Non-Adherence category. If you wish that the total number of calls in both
categories will sum up to the total number of calls that pass the filter, you need to set
the View contacts ranking from: for each of the categories so that they sum up to 6. For
example: 2 stars for the Adherence and 4 stars for the Non-Adherence.
Use Case Example

In a Call Center the manager wants to evaluate agents' adherence to a specific script. To
make such an evaluation, the Call Center manager can do one of the following:
 Create a Script Adherence percentage graph as follows:
a. Define a Script Adherence category or a Script Non-Adherence category called
Script-Privacy Act or Inverse-Privacy Act respectively.
b. In the Contact Analysis home page, define the same filter that was defined for
the above category from the Set Filters window.
c. In the Category chart, click on the bar that corresponds to the category you
defined above.

Chapter 7 - Category Definition Selecting Category Colors
d. Display the Agent chart in % mode.

For example, on average one agent appears to use the script 90% of the time
and another agent appears to use it only 20% of the time.
If the selected category is a Script Adherence category, the chart displays for
each agent the % of calls (out of those that passed the filter), in which the script
was identified.
If the selected category is a Script Non-Adherence category, the chart displays
for each agent the % of calls (out of those that passed the filter), in which the
script was not identified.
 Define a Script Non-Adherence category and perform a search to retrieve a list of
calls (including agent information) where the specific script was not used.
To save the resulting call set, select Export Contact Data (Excel) using the Create
Report feature.
 Change the state of a Script Adherence category or a Script Non-Adherence
category to Published in order to export its data to QM for further analysis.
Selecting Category Colors

The category color is consistently applied throughout the application and serves as a
visual aid for identifying the category in various areas of the application.
There are 16 available colors. Using these colors, you can create color schemes. For
example, the blue color can be used for all issues related to your products, and the
green color can be used for order processing issues.
The color of the Emotions category cannot be changed.
You can identify the category by color in the following locations:
 The Categories area in the home page
 The Category list in the Category Builder.
For example:
 The Category chart, where each bar is colored according to the color of the category
it represents.
For example:
 The Speech Analytics player. Categories to which the contact belongs are indicated
by their respective colcor.
For example:

Chapter 7 - Category Definition Defining the $ Impact of a Category
 The Transcription area. Terms that belong to a specific category are give their
respective category color.
For example:
NOTE In trends, the color of the category's line chart is automatically selected and is not
consistent with the color assigned to the category in the Category Builder.
To select the color for your category:

1 Click the Category Builder button to open the Category Builder.
2 In the Category column of the list, click the relevant name. The category definition
page opens.
3 In the category definition page, in the Category Color list, select a color.
Defining the $ Impact of a Category

A dollar impact value determines the potential value of contact segments belonging to
the category. For example, if problems in your web site cost your company more than
problems in order processing, you can assign the Web Site category the dollar value of
$200, and the Orders category the value of $100.
If on a given day 50 contacts are found in the Web Site category, and 60 in the Orders
category, the potential dollar impact for your company is 200x50=$10,000 for Web Site,
and 100x60=$6,000 for Orders, suggesting it is still more profitable for you to address
your Web Site issues before resolving problems in order processing. This value is used in
the Category by impact ($) chart. See “Using Charts” on page 127.

Chapter 7 - Category Definition Category Terms
To define the $ Impact value of a category:

1 Open the Category Builder. See “Opening the Category Builder” on page 165.
NOTE If the category's definition page is already opened, start the following procedure
from step 3.

page opens.
3 In the category definition page, in the $ Impact field, enter the required value.
Category Terms
This section outlines the information required for understanding how to work with
Category Terms, by covering the following topics:
 Category Terms, page 179
 Category Term Definition: Classic vs. Graphical View, page 181
 Switching Between Classic and Graphical View, page 182
 Defining Category Terms in Classic View, page 182
 Graphical View, page 185
 Enter a Term Correctly, page 194
 Warning Icons, page 194
Category Terms
Category terms are words, phrases, or combinations of them that are representative of
a certain business issue in customer calls. Once saved as part of a category definition,
terms serve to identify contacts pertaining to that issue and to qualify them for that
category – the Speech Analytics application automatically assigns a contact to a
category upon detecting that category’s terms in the contact (provided the contact
meets other parameters of the category definition, such as ranking and filters).
Terms are a central part of the category definition. A good selection of terms for a
category ensures that the category is wide enough to detect a maximal number of
contacts related to a specific issue, yet homogenous so as to exclude any irrelevant
contacts where the same terms can be used in other context.

The following are some examples of category terms:

 Contact segments using terms such as angry, unhappy, disappointed and
complain could be gathered together in a category called Complaints.
 Contact segments using terms such as return, damaged, cancel and never
received could be gathered together in a category called Orders. In this way, the
most common ordering problems can be identified and solved.
 Terms such as called NEAR back, called before, called earlier could identify
repeat calls. These terms can be assigned to the Repeat Calls category.
 Contact segments with the term C:"thank you" (where the customer thanks the
agent) could belong to the Customer Satisfaction category.
This section outlines the information required for understanding Category Terms, by
 Term Syntax, page 180
 Significance Level of Terms, page 180
 General Guidelines, page 181
 Where Are Category Terms Defined?, page 181
Term Syntax
A term can be a single keywords, a phrase enclosed in quotation marks, or an
expressions with operators.
The syntax of category terms is similar to the syntax of text queries that you submit
when performing searches. For a detailed description of all options of defining category
term, see “Building Text Queries” on page 70.
Significance Level of Terms

Each term must be assigned one of the significance levels:
 Very Important Term
 Important Terms
 Avoid Terms
Adding Avoid Terms can significantly reduce a category size. Contacts containing any of
the Avoid Terms cannot qualify for the category even if they contain Important Terms or
Very Important Terms.
TIP If you need to exclude a certain term, consider using the NOTIN operator with the
term and adding the term to Very Important Terms or Important Terms. For
example, you may need to find contacts containing the word "credit" but not when
it is used as part of the phrase “credit card”. Adding "credit card" to Avoid Terms
can filter out a great number of contacts. Instead, define your term as credit
NOTIN "credit card" and add it to Very Important or Important Terms.

See “Defining the Rank Threshold of a Category” on page 195 to understand the impact
of terms on the rank of segments in a category.
General Guidelines
 There is a maximum limit of 30 terms per category. However, most categories will
contain approximately 10-20 terms. Categories containing too many terms will be
too general and will include contacts not related to the category subject.
 Each term in a category has an OR relationship with the other terms. This means
that it is sufficient for a contact to contain one of a category's term to be qualified
for that category.
 Depending on the terms used, a contact can be assigned to one or more categories,
or have no category affiliation at all.
 A typical term should be 1-3 words long.
 Try to avoid using very short or one-syllable words.
Where Are Category Terms Defined?

Category terms can be defined in the Classic View or Graphical View tab of the Category
Definition page. To learn more about each view, see “Category Term Definition: Classic
Category Term Definition: Classic vs. Graphical View

There are two ways to define category terms in Speech Analytics:
 Using the Classic View tab
 Using the Graphical View tab
The Classic View allows you to view and manage category terms in three lists, according
to their significance level - Very Important Terms, Important Terms, Avoid Terms. In the
Graphical View, the terms are shown in a tree that gives a unified access to diverse
types of information required to comprehend how well a term serves (or will serve) a
category.
Both views share some functionality, including:
 Receiving context-based suggestions for terms.
 Retrieving and playing back contacts that contain existing and candidate category
terms.
 Ability to add terms to the category and assign them the required significance level.

In comparison with the Classical View, the Graphical View provides more support for
your judgment calls, making them better based and easier. Unique functions of the
Graphical View include:
 Graphical presentation of various contexts in the scope of the category.
 Visual and numeric indication of the terms' strength in the category.
 Ability to estimate any term's impact on the category's size.
 Drill-down to contact segments of a single term or a selection of related terms.
In Graphical View, these functions are provided not only for existing terms but for
candidate terms as well, which gives you means to estimate candidate terms without
adding them to the category.
See “Understanding the Graphical View” on page 185 to learn more about the category
term tree, its various elements, and the additional information they provide.
To learn about defining category terms in the Classic View, see “Defining Category
Terms in Classic View” on page 182. To learn about defining category terms in the
Graphical View, see “Defining Category Terms in Graphical View” on page 188.
Both views are synchronized - all operations performed in the Classic View are reflected
in the Graphical View and vice versa, so you can switch between the views at any time
during the category definition. See “Switching Between Classic and Graphical View” on
page 182.
Switching Between Classic and Graphical View

The Classic and Graphical Views are synchronized - all operations performed in the
Classic View are reflected in the Graphical View and vice versa. This allows you to switch
between the views at any time during the category definition, without losing any work
done on the category's terms.
In the category definition page, click the tab of the other view.
For example, when working in the Classic View, click the Graphical View tab to
display the category's terms in the tree in the Graphical View.
NOTE During transition from the Classic View to the Graphical View, the last operation
that you performed in the Classical View - either Suggest for Term or Suggest
for Category - is executed. The suggested terms are placed on tree in the
Graphical View, and their size and strength are calculated.
To open the Graphical View tab as a separate large window, click the maximize button
in the top right corner.
To minimize the Graphical View window, click the minimize button in the top right
corner of the window.

Defining Category Terms in Classic View

Category terms can be defined in the Classic or Graphical View. The definition in the
Classic View is described below. To learn more about the Graphical View, see “Category
Term Definition: Classic vs. Graphical View” on page 181.
NOTE Defining category terms requires thorough understanding of the terms' role in the
category. To learn more about category terms and to receive general guidance on
defining them, see “Category Terms” on page 179.
1 In the category definition page, make sure the Classic View tab is displayed.
2 In the My Term area, enter the term you want to add to the category. See “Enter a
Term Correctly” on page 194.
3 Click Suggest for Term to view the following:
 A list of context-based suggestions for the term, displayed under the My Term
area. See “Working with Context-Based Suggestions in a List” on page 81 for a
detailed description of this area.
 A list of contacts containing the term in the lower area of the window. You can
play any of the retrieved contacts. Use those search results to determine
whether the term brings useful results or whether it needs adjusting or
replacing.
 You can now click on one of the suggested terms to display it in the My Term
field instead of the original term. To view a list of contacts containing the
suggested term, click the Suggest for Term button again.
4 When you are satisfied with the term, define whether it has a high or low level of
significance for the category, and then add it to the relevant term group appearing
on the right side of the My Term area. For instance, the most important terms
should be added to the Very Important Terms group. Terms of lesser importance
should be added to the Important Terms group. Terms that you do not want
included in the search results should be added to the Avoid Terms group.
To add the term to the group, click the Add button next to the term group to which
it should belong.
5 Repeat steps 2 to 5 to add additional terms to the category definition as needed.
6 Once you have added all the terms that you can think of, you can use the Speech
Analytics application's suggestion features to find additional terms to add to your
category, to enhance the category effectiveness. Click Suggest for Category to
view the following:
 A list of context-based suggestions for terms included in the category, displayed
under the My Term area. See “Working with Context-Based Suggestions in a
List” on page 81.
 A list of all contacts that meet the category's definition in the lower area of the
window. These are the contacts that contain one of the Very Important Terms or
Important Terms list but do not contain any of the terms appearing in the Avoid
Terms list. You can play any of the retrieved contacts. Use those search results

to determine whether the term brings useful results or whether it needs

adjusting or replacing.
You can now click on one of the suggested terms to display it in the My Term
field. To view a list of contacts that contain that suggested term, click the
Suggest for Term button again.
If you are satisfied with the results, you can click Add to add the term to the
category, to the relevant term group appearing on the right side of the My Term
area.
If you now click the Suggest for Category button again, the list of retrieved
contacts will include all the contacts that meet the category definition, with the
new term included. A list of new term suggestions for the current category
definition will be displayed as well.
NOTE If you make any changes in the category's terms after clicking the Suggest for
Category button, the warning icons appear in the category definition page: in the
Suggested Terms area and in the Contact Segment list. To learn more about
the warnings, see “Warning Icons” on page 194.
7 When you are satisfied with the category terms, save the category.
The following is an example of the Customer Satisfaction category definition, with its
terms, in the Classic View.

Graphical View
This section outlines the information required for understanding how to work with the
Graphical View, by covering the following topics:
 Understanding the Graphical View, page 185
 Understanding Terms’ Strength and Size, page 187
 Defining Category Terms in Graphical View, page 188
 Testing and Adding Terms in Graphical View, page 190
 Drilling Down to Contacts from Graphical View, page 191
 Changing the Significance Level of Terms, page 192
 Removing Category Terms in Graphical View, page 193
Understanding the Graphical View

The Graphical View displays existing and candidate terms in a term tree that shows
connections between the terms and isolates different contexts in which the category
terms have been used.
The following is an example of the term tree in the Graphical View.

NOTE If you are creating a new category, the Graphical View opens empty, and the
category term tree is built in the course of the category definition.
If you open an existing category definition, the Graphical View displays terms that
have been added to the category and also automatically offers context-based
suggestions for the category. The list of contact segments in the lower area of the
page displays segments that belong to the category. For more details on the list of
retrieved contacts, see “Understanding the Category Definition Page” on page 168.
Main Elements
The main elements of the tree in the Graphical View are nodes, branches and terms. For
more information on these elements, see “Understanding the Term Tree Structure” on
page 86.
Tree Layout
The layout of the tree and the distribution of terms among its branches are often
indicative of the quality of your category definition. The more consolidated the terms are
(for example, when the category terms appear on one branch or on a few distinct
branches that are close to each other), the more focused the category is. And on the
contrary, if the terms are dispersed, the category definition is not focused enough.
Contact segments that qualify for such a category will not always be relevant.
Term Types
The appearance of each term in the tree serves as a visual indication of its role in the
category. The following term types can appear in the tree:
 Existing terms: terms that have been added to the category. These terms are
colored green (for example, current bill). An icon to the left of the term indicates
the significance level of the term:
- Very Important Term
- Important Term
- Avoid Term
 Candidate terms: terms that are displayed in the tree but have not been added to
the category. These terms are colored red (for example, current charges).
 Tested term: a single term that you are testing at the moment. The tested term is
displayed in bold dark-grey color (for example, amount). For more information on
testing terms, see “Testing and Adding Terms in Graphical View” on page 190.
 Current Dataset: the distance between the Current Dataset branch and every
other branch in the Graphical View, indicates the relevance of each branch (word) to
the set of contacts you retrieved (that is, the dataset).
NOTE The Current Dataset is not relevant when context-based suggestions are executed
from a category.

The font size of each term reflects that term's strength in the category - smaller font is
used for less dominant terms and bigger font for more dominant terms. You can also see
a term's strength in a tooltip. For more details, see “Understanding Terms’ Strength and
Size” on page 187.
For example, from the appearance of the term below, you can learn the following:
 This term is colored green, which means it is an existing category term.

 The icon to the left of the term ( ) shows that this term has been added as a Very
Important Term.
 To understand the strength of this term in the category, compare its font size with
font sizes of other terms.
Working in the Graphical View

 The recommended desktop resolution for working in the Graphical View is
1280x1024. However, the Graphical View can operate under the 1024x768
resolution.
 All operations in the Graphical View are available from left-click menus. Right-click
menus are reserved for the Internet Explorer operations.
Understanding Terms’ Strength and Size

Term Strength
Term strength is one of the most important parameters you should take into account
when defining category terms. The strength is based on the percentage of the
category's top contacts in which the term is used. The higher is the percentage of
contacts where the term is used, the stronger is the term. The strength is ranked from 0
to 5 stars whereby a 5-star rank corresponds to the strongest terms in that category.
The Speech Analytics application determines the strength of all terms appearing in the
Graphical View - both existing and candidate ones. For existing terms, the number
shown is the actual strength of the term. For candidate terms, the strength is
determined as it would be if the term were added to the category.
The term strength is indicated in two ways - by the font size and in the tooltip.
 The font size serves as a visual indication of the term strength - the bigger is the
font, the stronger the term is. For a more detailed explanation of the terms' font
size, see the Term Types section in “Understanding the Graphical View” on
page 185.
 The tooltip shows the strength, ranked from 0 to 5 stars.
To view a tooltip on a term:

 Hover your mouse over the term.
The following are tooltip examples.

Notice the correlation between the terms' strength and the font size. In the left
example, the strength of the term "charge", which is displayed in a bigger font, is 5
stars. In the example on the right side, the term's font size is quite small, and the term's
strength is smaller as well (3 stars). Alongside with the strength, its related parameter,
size, is calculated for each candidate term.
A term's strength cannot be determined if the number of retrieved contacts is not
sufficient for the calculation. The minimal number of contacts required is 100. If a
category has fewer than 100 contacts, the strength of existing terms cannot be
calculated. Yet, if a candidate term can increase the category's size up to 100 contacts,
its strength is calculated.
Term Impact on the Category Size

The size of each term shows an estimate of how adding this term as an Important Term
would increase the number of contacts for this category, in percent. For example, if the
indicated size of a term is 20.00%, then adding it to a category containing 1,000
contacts would increase the category's size to 1,200.
The size of a term is shown for candidate terms only, and you can view it in a term's
tooltip, as described above (see “To view a tooltip on a term:” on page 187). If the
strength of a certain term cannot be calculated, the size is not calculated either.
TIP To estimate the impact of a term on a category size in case you add it to Avoid
Terms, try adding this term to the category and then review the number of
contacts retrieved. The number should be lower than the previous number,
because adding a term to Avoid Terms decreases the category size.

Defining Category Terms in Graphical View

Category terms can be defined in the Classic or Graphical View. The following is a typical
flow for defining category terms in the Graphical View. To learn more about the Classic
View, see “Category Term Definition: Classic vs. Graphical View” on page 181.
NOTE Defining category terms requires thorough understanding of the terms' role in the
category. To learn more about category terms and to receive general guidance on
defining them, see “Category Terms” on page 179.
1 In the category definition page, make sure the Classic View tab is displayed.
2 Determine a "seed" term for your category. The term should be representative of
the category's subject and likely to be found in calls related to that issue.
3 Enter the term into the field in the My Term area. See “Enter a Term Correctly” on
page 194.
4 Click the Suggest for Term button. A list of context-based suggestions is displayed
in the Suggested Terms area. The term and suggestions are also displayed in the
tree. At this stage, all terms appear in the same font size, because the category's
context has not been defined and therefore the terms' strength cannot be
calculated. The list of contacts containing the term is displayed in the lower area of
the page.
5 Review how the suggestions are distributed among contexts. Typically, each branch
represents a separate context.
6 Add a few terms to the category by left-clicking each term and selecting the option
with the required significance level.
7 Click Suggest for Category to refresh the tree and to have the terms' strength
calculated. In the refreshed tree, the terms are displayed in different font sizes,
corresponding to their strength.
8 Test terms and add the relevant ones to the category.
While testing terms, focus on stronger terms, but also use your judgement and
knowledge of the business issue to determine which terms are relevant. Strong
terms do not necessarily need to be added to the category, for instance, if it is a
common word such as "yes", or if the term is not related to the category context.
For more information, see “Testing and Adding Terms in Graphical View” on
page 190.
Category button or after moving from the Classic View to the Graphical View, the
warning icons appear in the category definition page: in the Suggested Terms
area, in the Contact Segment list, and in the Graphical View. To learn more
about the warnings, see “Warning Icons” on page 194.
9 As you add terms to your category, try to make your category as homogenous as
possible. The tree layout should clearly present one or, possibly a few, distinct
branches related to the same context. On the other hand, if your category terms are
distributed through the entire tree, then more work may be needed on the terms.

10 When you are satisfied with the category terms, save the category.
TIP Test your category definition with a term that is not related to the category
context. For that, in the My Term area, enter the unrelated term and click the
Suggest for Term button to see how the term will be placed in the tree. The term
should appear on a separate branch of the tree, clearly distinguishable from the
category's branch.
Testing and Adding Terms in Graphical View

When defining category terms in the Graphical View, you can first test candidate terms
in the tree and then add only those which you consider relevant for your category. The
testing methods available in the Graphical View include:
 Viewing a term's strength and its potential impact on the category's size.
 Retrieving and playing contact segments containing the term.
 Viewing the changes that this term can cause in the category definition: the
resulting category size, the re-calculated strength of exiting terms, and distribution
of terms by context.
If the term you wish to test does not appear in the tree, you can introduce it yourself.
To introduce a new candidate term:

1 With the Graphical View tab displayed, enter the term you want to test in the field in
the My Term area. See “Enter a Term Correctly” on page 194.
2 Click the Suggest for Term button. The term is added to the tree in the Graphical
View, where it appears in the bold grey font. In addition, the following changes
occur:
 In the Suggested Terms area, a list of context-based term suggestions is
displayed.
 In the lower area of the window, a list of contact segments containing the term
is displayed.
To test and add a candidate term:

 In the tree view, left-click the term and select Suggest for Term.
or
 Select a term from the list of context-based suggestions and click the Suggest
for Term button.
The following changes occur:
- The term is displayed in the bold dark grey font.
- In the Suggested Terms area, a list of context-based term suggestions is
displayed.

- In the lower area of the window, a list of contact segments containing the term
is displayed.
2 Analyze the term:
e. Review the term's strength and its potential impact on the category's size. For
more information, see “Understanding Terms’ Strength and Size” on page 187.
f. Play retrieved contact segments to better understand how relevant the term is
for the category.
3 Try adding the term to the category to estimate its impact. This operation does not
change the category definition yet, but results in recalculating other terms' strength
and their impact on the category size.
a. Left-click the term and select to add it with the required significance level: as a
Very Important Term or Important Term. An icon to the left of the term
appears indicating the selected significance level.
NOTE Avoid Terms can be added in the Classic View only. For more details, see “Defining
Category Terms in Classic View” on page 182.
b. To refresh the tree, left-click the term and select Suggest for Term.
Alternatively, you can click the Suggest for Category button, or the Suggest
for Term button with a term appearing in the My Term field.
c. Review how other terms' strength and their impact on the category size have
changed because of the new term.
Category button or moving from the Classic View to Graphical View, the warning
icons appear in the category definition page: in the Suggested Terms area,
in the Contact Segment list, and in the Graphical View. To learn more about the
warnings, see “Warning Icons” on page 194.
4 If you are satisfied with the term, click Save Category. The category is saved.
5 If you do not want to add the term to the category, continue working without saving
your changes. The term will be removed from the tree with the next operation you
perform in the Graphical View (for instance, when you drill down to contact
segments of a different term in the tree).
Drilling Down to Contacts from Graphical View

When working in the Graphical View, you can drill to contacts containing terms displayed
in the tree. Drilling to contacts does not affect the category definition, it only retrieves
contacts in which the terms have been used.

The ability to drill to contacts is available on different elements of the tree. Depending
on the selected element, one or more terms can be included when drilling to contacts.
You can drill to contacts on the following elements:
 On a single term, to retrieve contacts where that term has been used. For example:
This drill-down will retrieve contacts containing the term "being charged".
 On a leaf node, to retrieve contacts that contain all terms of that leaf node. For
example: This drill-down will retrieve contacts containing terms on the selected leaf
node.
 On any non-leaf node, to retrieve contacts with terms of all leaf-nodes of the
branch. For example: This drill-down will retrieve contacts containing terms on four
leaf nodes of this branch.
Each term included in the drill to contacts has an OR relationship with the other terms.
This means that it is sufficient for a contact to contain just one of the terms to be
retrieved. Therefore, a list of retrieved contacts grows with the number of terms to
which you are drilling. Yet, the retrieved contacts are focused around the same issue,
because terms included in the drill to contacts in the Graphical View are always related
to each other. The strongest correlation exists among terms of the same leaf node.
Terms on different leaf nodes of the same branch can be less correlated but typically
belong to same context. For more information on different elements of the tree, see
“Understanding the Term Tree Structure” on page 86.
NOTE Drilling to Avoid Terms cannot retrieve any results. Speech Analytics prevents you
from performing such operations and issues a message requesting to include at
least one term of a different or no significance level into your drill to contacts.
To drill to contacts from the Graphical View:

 In the Category Definition > Graphical View tab, left-click a term, a leaf node, a
middle node and select Drill to contacts.
A list of contacts containing the terms to which you drilled down is displayed in the
lower area of the page.
Changing the Significance Level of Terms

To change the significance level of any category term.
1 In the Graphical View, left-click the term.
2 Select the available option under the Move to entry:
 For an Important Term, the available option is Move to > Very Important
Terms.

 For a Very Important Term, the available option is Move to > Important
Terms.
The icon next to the term changes according to the selected significance level.
If you make any changes in the category's terms after clicking the Suggest for
3 Left-click the term and select Suggest for Term to refresh the tree. Alternatively,
you can click the Suggest for Category button, or the Suggest for Term button
with a term appearing in the My Term field. The tree is updated according to the
change: the tree layout changes, and the strength of all terms is recalculated.
4 If you are satisfied with the results, click Save Category.
If you do not save your change, the previous significance will be restored with the
next operation you perform in the Graphical View.
5 Click Save Category to save your change.
Removing Category Terms in Graphical View

You can remove any term that has been saved as part of the category definition.
1 In the Graphical View, left-click the term and choose Remove from ..... (Very
Important Terms, Important Terms, or Avoid Terms). The term is removed
from the tree.
If you make any changes in the category's terms after clicking the Suggest for
2 Click the Suggest for Category button to refresh the tree. The tree is updated
according to the change: the tree layout changes, and the strength of all terms is
recalculated.
3 Review the changes caused by the removal of the term.
4 If you are satisfied with the results, click Save Category.
If you do not save your change, the term will be restored with the next operation
you perform in the Graphical View.

Enter a Term Correctly

When entering a term in the My Term area, note the following:
 In the My Term area, you always work with one term at a time. If your term
contains several words, enclose them in quotation marks or connect with an
operator.
For example, if you want to add both the word angry and the word annoyed as
terms to a category, do not enter both of them into the My Term field
simultaneously, but enter each of them at a time. Alternatively, you can build an
expression with an operator to combine these words into one term, for example,
angry AND annoyed. See “Building Text Queries” on page 70s for a detailed
explanation on various options of defining category terms.
 By default, the autocompletion feature is enabled. Autocompletion helps identify
similar terms to the one you entered and also measures their frequency and
functions as a spelling aid.
With autocompletion, each time you start typing a new word, a list of possible word
completions matching the entered characters appears. The list displays words found
in transcribed contact segments. Choose any of the relevant words to use them as
part of your term.
To learn more about working with autocompletion, see “Using Autocompletion in
Text Queries” on page 77.
Warning Icons
The Category Definition page displays warning icons if, after clicking the Suggest
for Category button or moving from the Classic View to the Graphical View, you make
any of following changes in a category's definition:
 Change the category's terms: add new terms, change their significance level, or
remove terms.
 Change the category's filter: create a new filter, edit an existing filter, or clear it.
The warning icons inform you that because of your latest change both the category's
term suggestions and the list of retrieved contact segments have become irrelevant.
When you are working in the Classic View, the warning icon appears in the
Suggested Terms area and in the contact segment list.
For example, when you are working in the Graphical View, the warning icon appears
in the Suggested Terms area, in the contact segment list, and also in the Graphical View
tab. For example:
You can continue working with the warnings displayed. But if you want to view the
category's relevant term suggestions and contact segments, click the Suggest for
Category button or the Suggest for Term button (with a term displayed in the My
Term field). This will update suggested terms and retrieve relevant contact segments. If
you are working in the Graphical View, the term tree will be also updated: its layout will
be changed, terms in the tree will be updated according to the list, and the strength of
each term will be determined.

Chapter 7 - Category Definition Defining the Rank Threshold of a Category
Defining the Rank Threshold of a Category

Ranking is the process by which a contact segment is given an overall score for a
category according to the terms that the segment contains.
When determining a segment's rank in a category, Speech Analytics takes into account
the significance level of each term. For instance, segments containing Very Important
Terms will receive higher ranking than those with Important Terms only. The higher a
a segment's rank, the more relevant the segment is for that category.
A category's rank threshold determines the minimum rank a contact segment must have
in order for it to qualify for the category. The rank can range from 5 stars for the most
relevant contact segments to 1 star for the least relevant ones.
If the category contains both Very Important Terms and Important Terms, then contacts
that do not contain Very Important Terms cannot receive a 5-star rank. Yet, contacts
containing Important Terms only can receive a 5-star rank if no Very Important Terms
have been defined for the category.
To define the rank threshold of a category:

1 Open the Category Builder. A list of categories is displayed. See “Opening the
Category Builder” on page 165.
from step 3.
2 In the list, click the category's name, or click the Edit icon on the category's row.
The category's definition is displayed.
3 In View contacts ranking from, select the minimum rank a contact must have to
qualify for the category. The default is 3 stars.
Defining Category Filters

Filters can help you focus your category on a predefined set of calls - for example,
relating to your business issue or handled by a specific group of people. If a category
includes a filter, then only those contact segments that match both the filter and the
category term definition can qualify for that category.
This section outlines the information required for understanding how to define category
filters, by covering the following topics:
 Hit or No Hit Segments, page 196
 The Impact of Category Filters on Category Performance in Trends, page 196
 Previewing Category Filters, page 196
 How to Define a Category’s Filter, page 197

Chapter 7 - Category Definition Defining Category Filters
Hit or No Hit Segments

With regard to category filters, two types of contact segments are distinguished: "hit"
and "no hit" segments. "Hit" segments are segments that belong to a category as
matching both its filter and its term definition. "No hit" segments match the category
filter only but do not match the term definition, and therefore do not belong to the
category.
For example, a category's filter can limit the minimum duration of a contact segment to
3 minutes. With that filter:
 The category's "hit" segments are all recordings that are 3 minutes long or longer
that match the category's term definition.
 "No hit" segments are the segments that are 3 minutes long or longer but do not
match the category term definition.
Now let's assume Speech Analytics processed 1,000 contact segments on a given date,
and 900 segments matched the filter. If 300 out of those 900 segments also matched
the category's terms definition, then there are 300 "hit" segments and 600 "no hit"
segments for the category on that day.
NOTE "Hit" and "no hit" data can also be used in systems integrated with Scorecards.
With the integration, daily aggregated data of "hit" and "no hit" segments becomes
available in Scorecards, where it can be used for creating KPIs that measure agent
performance. For more information on creating KPIs in Scorecards, See the
Scorecards Configuration and Maintenance User Guide. For information on how to
interpret KPI data shown in Scorecards, see the Scorecards User Guide.
The Impact of Category Filters on Category Performance in

Trends
A category filter has an effect on the category's performance calculation in trends. A
category's performance in trends is the percentage of "hit" segments out of the total
number of segments that match the category's filter ("hit" + "no hit" segments). For
more details on calculating category performance, see “Category Trends” on page 25.
Previewing Category Filters

You can view whether a category's filter has been set in the category list or in the
category definition page.
To check a category's filter in the category list:

1 In the Category list, locate the category and see the values displayed for this
category in the Terms column.
2 If a category includes filters, or contains multiple keywords whose list is too long to
fit one line, the cell appears collapsed, with a plus button at the beginning and an
ellipsis (...) at the end.

Chapter 7 - Category Definition Defining Category Filters
3 Click the plus button to view the entire content of the cell. If a filter has been set,
its definition appears after the terms, at the bottom of the cell.
For more details on the category list, see “Understanding the Category List” on
page 166.
To check a category's filter in the category definition page:

1 Read the text appearing in the top left corner of the category definition page, in the
Filter area. If the category has no filter, the text reads "Category filter not applied".
Otherwise, if the category has a filter, the text reads "Category filter applied".
2 If the category has a filter, hover the mouse over the area to preview the filters in a
tooltip.
For more details on the category definition page, see “Understanding the Category
Definition Page” on page 168.
How to Define a Category’s Filter

It is not mandatory to define category filters. You can add a category's filter when
creating the category, or you can create categories without filters and add the filters at
later time. You can also edit or clear any defined filter of any category.
To define filters for a Speech Analytics category:

from step 3.
page opens.

Chapter 7 - Category Definition Defining Category States
3 In Category Definition page, in the Classic View tab, click the Define Filter
button.
The Filters window is displayed showing the previously set filters.
4 Define filter values in one or more of the available fields.
NOTE If the default values in the Date Range filter are not modified, All dates is
applied to the category when it is saved.
If you would like to limit the category to a specific date range, select one of the
Date Range parameters, enter a value click Save.
5 Click OK.
If you change your mind while defining the filter, click Clear.
6 When you finish applying filters, click Close.
NOTE If in the course of a category definition you clicked the Suggest for Category
button and then made any changes in the category's filter, the warning icons
are displayed in the category definition page. To learn more about the warnings,
see “Warning Icons” on page 194.
Defining Category States

Each category is assigned a certain state that determines the category’s role and
availability in Speech Analytics and in integrated applications. The initial state of the
category is defined when the category is created. You can change a category's state at
any later time.
This section outlines the information required for understanding how to define category
state, by covering the following topics:
 Category States, page 198
 The Impact of Changing a Category’s State, page 200
 How to Define a Category’s State, page 201
Category States
A category can be in one of the following states:
 Published - Published categories are used for automatic contact categorization,
searches, and TellMeWhy™ analysis. Their data can be accessed in other integrated
applications such as Quality Monitoring, Reports and Data Analytics. As such,
published categories can be used as search criteria in the Portal, as filters in
Reports, and as performance metrics and elements of instance rules in the Data
Analytics application. Published categories are updated on an automatic basis when
the Published category is synchronized (see “Synchronizing Categories” on
page 203).

Chapter 7 - Category Definition Number of Categories
NOTE The Emotions category will only be available for contact segments associated with
instances where this category is published.
 Active - Similarly to published categories, active categories are used for automatic
contact categorization, searches, and TellMeWhy™ analysis in Speech Analytics. Yet,
their data is available in this application only and cannot be accessed from other
applications. For example, an active category cannot be used as a search criterion in
Quality Monitoring. Contact segments belonging to an active category in Speech
Analytics have no category affiliation in other applications.
Both published and active categories are displayed in the Speech Analytics home
page, and their trends and trend analysis data are calculated.
 Inactive - Inactive categories are the categories which are not in use at the
moment. They can serve as drafts for future categories or represent outdated
categories which are no longer relevant but which you want to keep. Inactive
categories can be viewed in the Category Builder only.
Inactive categories are disregarded during the categorization process. Even if a
contact segment qualifies for an inactive category, Speech Analytics does not tag
that contact with the category. Inactive categories cannot be used for searches and
analysis - neither in Speech Analytics, nor in integrated applications.
Each Speech Analytics instance can have a specific amount of categories. For more
information, refer to “Number of Categories” on page 199.
Number of Categories
The total number of categories in one Speech Analytics instance depends on the type of
server on which the Speech Analytics Application role is deployed.
 When working with a Standalone Speech Analytics Application server, the total
number of categories must not exceed 500.
- Out of the 500 categories you can have a maximum of 100 Active categories.
- Out of the 100 Active categories, a maximum of 100 can be Published
categories.
That is, the combination of Active and Published categories cannot exceed 100.
For example, if 60 categories are Active and 40 categories are Published, there can
be up to 440 Inactive categories.
 When working with a Speech Analytics Application Role on the Consolidated server,
the total number of categories must not exceed 100.
- Out of the 100 categories you can have a maximum of 50 Active categories.
- Out of the 50 Active categories, a maximum of 50 can be Published categories.
That is, the combination of Active and Published categories cannot exceed 50.
For example, if 40 categories are Active and 20 categories are Published, there can
be up to 60 Inactive categories.

Chapter 7 - Category Definition Number of Categories
The Impact of Changing a Category’s State

The following are some important implications of changing a category's state:
Published Categories
Only categories that have a Published state are used and accessed by other suite-wide
applications in the system (such as Quality Monitoring and Scorecards). Therefore, when
you change a category's state from another state to Published, its category results are
integrated with these suite-wide applications, and when you change a category's state
from Published to another (unpublished) state, the category results are no longer
integrated with these applications.
In addition, when you modify specific fields for a Published category, the system
synchronizes the Published categories results with system databases, and these updates
are reflected in the speech category data used and accessed by the other suite-wide
applications.
Modifying any of the following parameters of a category definition will trigger the
category results of the category to be synchronized with suite-wide applications:
 Rank
 State
 Very important terms
 Important term
 Avoid terms
 Define Filter
The synchronization of the Published category result set may take up to one hour to
complete.
For additional information about category synchronization, see “Synchronizing
Categories” on page 203.
If you change the state of a published category to Active, incoming contact segments
continue to be tagged for this category. However, this information will only be available
in Speech Analytics. In other applications, these contact segments will have no category
affiliation in other applications.
Activating Categories
You can activate any category by changing its state from inactive to active or published.
Upon the change, the category will be retroactively assigned to all relevant contact
segments in the Speech Analytics online database.
Deactivating Categories
You can deactivate a category by changing its state to inactive. None of the incoming
contact segments will be tagged with this category. All segments in the Speech Analytics
online database that were tagged with this category will lose their tag. If an active or
published category was deactivated, and then activated again later, there will be a gap

Chapter 7 - Category Definition Copying Categories
in the historical trend of the category for the period of time when the category was
deactivated.
How to Define a Category’s State

You define the initial state of a category when creating it. You can change this state at
any later time. The state can be changed to any of the available values, regardless of
the category's present state.
If you are creating a new category, perform step 3 of the following procedure and then
return to “Creating Speech Analytics Categories” on page 170.
TIP You can view the current state of the category in the State column of the category
list. For more details, refer to Understanding the Category List.
To define a category’s state:

from step 3.
page opens.
3 In the category definition page, in the State field, select the required state.
4 Click the Save Category button to save the change.
Copying Categories
1 From the Category Builder, open the category that you want to copy by clicking the
category's name or clicking the Edit categories icon .
2 If necessary, make any changes in the category's definition.
3 Click Save Category As.
4 Enter a new name for the category and click OK. The category is saved under the
new name.

Chapter 7 - Category Definition Editing Categories
Editing Categories
When you edit a category's terms, rank threshold or filter, the following changes occur in
the application:
 All contact segments available in the Speech Analytics online database at the
moment are scanned, and the relevant ones are retroactively assigned to the
category. Contact segments that are no longer relevant for the category are
disassociated from it. The number of contact segments for the category in the
Contact Analysis tab is updated. For more details, see “Contact Analysis Tab” on
page 64.
database.
 The entire Index data trend of the category changes to reflect the category's
performance according to its new definition. For more details, see “Adding Index
Data Category Trends in the Plot” on page 34.
 If the change in the category's definition has caused a significant increase or
decrease in its performance, the category's historical trend changes accordingly on
the date of modification. After that date the historical trend reflects the category's
performance according to its new definition. For more details, see “Adding Historical
Category Trends to the Plot” on page 33.
You can edit a saved category, including the Emotions category.
To edit a category:
2 In the list, click the category's name, or click the Edit icon on the category's row.
The category's definition is displayed.
3 Change any of the following properties of the category:
 “Selecting Category Colors” on page 177
 “Defining the $ Impact of a Category” on page 178
 “Category Terms” on page 179
 “Defining the Rank Threshold of a Category” on page 195
 “Defining Category Filters” on page 195
 “Defining Category States” on page 198
4 If, at any point in the category definition process, you decide to cancel the changes
you have made to an existing category, click Reset Category in order to revert
back to the state the category definition was in when you entered it to make
changes (that is, the last time the definition was saved).

Chapter 7 - Category Definition Synchronizing Categories
5 Click Save Category to save your changes. This button is enabled only when you
have made a change to a category.
TIP Always click the Back to Category List button to leave the
category definition page. Note that if you close the page by clicking the Close
button , any unsaved changes will be lost.
Synchronizing Categories
Speech Analytics category results are synchronized with system databases when specific
category updates are performed. Synchronizing Speech category data with system
databases allows other suite applications (such as Quality Monitoring and Scorecards) to
access and use updated Speech category results. As such, updated published categories
are used as search criteria in the Portal, as filters in Reports, and as performance
metrics and elements of instance rules in the Data Analytics application.
Speech Analytics category results can be synchronized with system databases when a
Published category is modified or any category is deleted. Users can decide whether to
synchronize the data as soon as the category modification is performed, or postpone the
synchronization for a later time.
An icon next to each category state indicates whether the category is in the process of
being synchronized, or whether an error occurred during the synchronization process:
 indicates that the category is in the process of synchronization.
 indicates that synchronization failed.
 indicates a problem with the connection to the database.
 indicates that the synchronization status is unknown due to an upgrade from a
previous version. That is, category results from a previous version are not
automatically synchrnonized in with the results from the current version.
The synchronization process is initiated as a result of one of the following scenarios:
 A Published category is modified
 An Inactive/Active category is changed to Published
 A new category is saved as Published
 A Published category is deleted
 When you modify the definition of a published or active category that is part of a
Published category's filter, the Published category is synchronized.
When the user clicks Save or Save Category As, upon completing one of these
scenarios, a confirmation message with the following two or three options is displayed:
 Save
To synchronize platform databases with the new category results, click Save.

Chapter 7 - Category Definition Deleting Categories
 Save as Active Category

To not update the database, click Save as Active Category. Active categories with no
dependent Published categories are not synchronized with the platform databases.
 Cancel
To disregard category changes, click Cancel.
Speech Analytics categories can be modified regardless of their synchronization status.
If a user modifies or deletes a category that is being synchronized, the change cancels
the active synchronization process with a new one.
Once the synchronization process is initiated, it can take up to one hour for all the
databases to be updated with the new category results.
Deleting Categories
When a category is deleted, it is no longer available in the application. In addition, the
following changes occur:
 The category is disassociated from all contact segments in the Speech Analytics
online database to which it was assigned.
 The category cannot be used for searching and filtering. If the category was used as
a filter in another category, the other category will be empty, with no contact
segments associated to it.
 The category's trends and their analysis are no longer available.
To delete a category:
2 In the list of categories, click the Delete icon next to the category that you want
to delete.
You cannot delete a category that is part of a Published category's filter and you cannot
delete the Emotions category.

Chapter 7 - Category Definition Exporting and Importing Categories
Exporting and Importing Categories

Speech Analytics enables you to import and export categories. You can use these
abilities for backup purposes or for sharing categories with other sites or Speech
Analytics instances.
When exporting categories, you can choose to export all or selected categories. The
export creates a separate file, by default categories.icd, that stores a list of exported
categories, with their definitions. During the export, you can save the file to any
location. Once the categories.icd file is saved, you can import it into any Speech
Analytics instance to recreate the categories that you exported.
For example, if you need to create the same categories at all sites of your organization,
you can create all the categories at one site, export the categories to a separate file, and
then import this file into the Speech Analytics at other sites.
A Speech Analytics instance can have up to 200 categories. Before importing categories,
always make sure that the resulting number of categories will not exceed that number. A
category import that may result in more than 200 categories cannot be performed. For
example, if you try to import 50 categories in addition to existing 170 categories, no
categories will be imported. In such cases, a corresponding message is issued to you by
Speech Analytics.
IMPORTANT Some category filter parameters may not be imported when importing categories.
For more information, see “Importing / Exporting Category Filters” on page 232
When importing categories:

 You can choose whether to import the Emotions category, the other categories or
both.
 If you currently have categories with the same name as the categories you want to
import, you can choose whether to overwrite the current categories, or rename the
categories you want to import. Importing the Emotions category always overwrites
the existing Emotions category.
 All categories are imported as inactive, regardless of their initial state. If required,
you can change the status of any imported category. See “Defining Category
States” on page 198 for more details on changing the category's status.
To export categories:
1 From the Category Builder, select the categories you want to export and then click
Export.
The Export Categories window appears.
2 If any categories were selected in the category list, additional option buttons are
displayed in the Export Categories window: All categories and Selected
categories. Leave the Selected categories option selected to confirm your intention
to export selected categories only, or select All categories to export all existing
categories.
3 Click OK.

Chapter 7 - Category Definition Exporting and Importing Categories
4 In the File Download window, click Save.

5 In the Save As window, choose the location and click Save.
6 Click Close. The file is saved as categories.icd.
To import categories:
1 From the Category Builder window, click the Import button.
2 Click Browse and select the location of the category definition file you want to
import. The file must have a .icd suffix. By default, the file name is categories.icd.
3 Click Open.
4 Select whether you want to import the Emotions category, all other categories or
both.
5 Click OK.
6 If the Speech Analytics application contains categories with the same name as the
categories you want to import, a Conflicts window appears listing all the conflicting
category names.
 Select Overwrite to overwrite the current category definitions and select OK to
confirm.
 Select Rename to rename the imported categories (and thus keep your current
category definitions).
7 An Import Results window appears showing either Modified if you selected the
Overwrite option or Renamed if you selected the Rename option.
8 Click Close.
All imported categories are added to the list of categories. The state of all imported
categories is inactive.
9 If needed, change the state of imported categories. For more details, See “Defining
Category States” on page 198.

Chapter 7 - Category Definition Creating Category Definition Reports
Creating Category Definition Reports

You can create a report with detailed category definitions, in PDF format, to print or
save.
1 In the Category Builder window, select the categories whose definitions you want
p
to include in the report and click the button
or
Click the Create Report button without selecting any categories to include all
category definitions in the report.
The Category Definition Report window opens.
2 Give the category definition report a unique name. This is the name that will appear
for this report on the Saved Reports page.
3 If any categories were selected in the category list, additional option buttons are
displayed in the Category Definition Report window: All categories and Selected
categories. Leave the Selected categories option selected to confirm your intention
to create the report on selected categories only, or select All categories to include all
existing categories in the report.
4 Select or clear the Open Saved Reports Screen check box. If you select the check
box, the Saved Report screen will be automatically opened immediately upon report
creation, and you will be able to access your report.
5 Click OK. If you selected the Open Saved Reports Screen check box, the Saved
Report window is displayed; otherwise, the Category Builder window is displayed.
NOTE Refer to “Managing Reports” on page 220 for more details on accessing saved
reports and working with them.

Chapter 8
This chapter outlines the information required for understanding how to work with
reports and how to export data, by covering the following topics:
 Reports and Data Export, page 209
 Creating Chart Reports, page 209
 Creating Contact Data Reports, page 213
 Compare Two Result Sets Overview, page 215
 Managing Reports, page 220
Chapter 8 - Reports and Data Export Reports and Data Export

The Speech Analytics application offers the following methods of data export for sharing
and backup:
 Creating Chart Reports, page 209
 Creating Contact Data Reports, page 213
 Compare Two Result Sets Overview, page 215
In addition, you can also export and import category data. For more details, refer to:
All export tasks and reports are listed in the Saved Reports window from which you can
open them, save to other locations, or delete once this information is not longer needed.
See “Managing Reports” on page 220.
Creating Chart Reports

You can export charts data to the Excel or PDF format by creating chart reports, which
you can then save to any location or print.
Both Excel and PDF reports contain the same header that includes:
 Report type and name.
 Name of the Speech Analytics instance where the report was created.
 Creation date.
 Details of the search query.
 Search filter.
 The number of retrieved contact segments and their percentage out of the total
number of segments.
 Query result metrics.
NOTE If values of Speaker Separation metrics are not available (Average Agent Talk-Time
Percentage, Average Customer Talk-Time Percentage, and Average Talk-Over-Time
Percentage), the report does not include any segments recorded with Speaker
Separation.
The body of the chart report depends on the selected report format. The PDF report only
displays the first page of the results received from the selected charts. The Excel report

Chapter 8 - Reports and Data Export Creating Chart Reports
shows each chart's data in a textual format, which can be more suitable for analysis.
The Excel chart reports include the following values per each chart included:
 Name - displays either Agent, ANI, Time of day, etc., depending on the type of
chart. For example, in a "Time of day" chart, the chart is divided between different
times of day (morning, evening, etc.) and these values will be displayed.
 Value - the number of contact segments corresponding to the specified query and
filter.
To create a report containing chart data:

1 From the Speech Analytics home page, or from the Results page, click the Create
Report button.
The Create Reports dialog box opens.
2 Select Reports.
3 Give the report a unique name. This is the name that will appear for this export task
on the Saved Reports page.
4 Select Export to PDF or Export to Excel.
NOTE For export to PDF, additional options are available:
 Display the charts with legends
 Display values as percentages - this option appears only if you
click on the Create Report button from the Results page.
Select the required option for exporting charts to the PDF format.
5 Click the check boxes beside the charts that you wish to include in your report, or
click Select All Charts to include all available charts.
6 Select or clear the Open Saved Reports Screen check box.
 If the check box is selected (default), the Saved Reports window is
automatically opened after you save the report. From the Saved Reports
window, you can open the report immediately upon its generation.
 If you clear the check box, then after you save the report, the Create Report
window will be closed, and you will return to the page from which you created
the report.
TIP You will be able to open the report from the Saved Reports window at a later time
by clicking the Saved Reports button in the Speech Analytics home page.
For information on the Saved Reports window and managing saved reports, see
“Managing Reports” on page 220.
7 Click OK to save the report.

Excel Report Example

PDF Report Example
NOTE If reports produced include single-term queries without the IDF calculation (which
is the default functionality), the report will indicate this exclusion by an asterisk
(*), along with a note at the bottom of the report indicating that the IDF
calculation was excluded. For more information on IDF, see “Ranking and
Document Relevancy” on page 223.

Chapter 8 - Reports and Data Export Creating Contact Data Reports
Creating Contact Data Reports

You can export contact data by creating a report containing certain contact segments.
The report can include all segments, a partial selection of segments (as chosen by you),
a random selection of segments (automatically selected) or a specific number of
contacts starting from the first contact.
Speech Analytics allows export of up to 10,000 contact segments and displays the
exported contact information in an Excel file.
Contact Data Report Example

Chapter 8 - Reports and Data Export Creating Contact Data Reports
The report's header includes:

 Report type and name.
 Name of the Speech Analytics instance where the report was created.
 Creation date.
 Details of the search query.
 Search filter.
 The number of retrieved contact segments and their percentage out of the total
number of segments.
 Query result metrics.
NOTE If the report does not include segments recorded with Speaker Separation, values
of metrics specific to Speaker Separation (average percentages of agent talk time,
customer talk time and talk-over time) are not available. For more information
Speaker Separation, see “Speaker Separation” on page 18.
The report body displays a line per each contact segment that includes:
 All the information available for that segment in the Result page, in the context of
that search. For example, segment's rank in the search and the keywords found.
 The categories to which the segment belongs.
 Contact details. For a detailed description of contact details included in the report,
see “Understanding Contact Details” on page 116.
To create a contact data report:

 If you wish to export all contact segments or their random selection, click the
Create Report button on the toolbar - either from the home page or from the
Results page.
 If you wish to export specific contact segments:
a. Perform a search.
b. From the Results page, select the contact segments to export.
c. Click the Create Report button on the toolbar.
The Create Reports dialog box opens.
2 Give the report a unique name. This is the name that will appear for this export task
on the Saved Reports page.
3 Select Export Contact Data.
4 Choose one of the following:
 All Contacts - to export all contact segments matching the current query and
filter. A limitation of 10,000 applies if there are more than 10,000 contacts in the
database.

Chapter 8 - Reports and Data Export Compare Two Result Sets Overview
 Selected only - This option is displayed only if you are creating the report from
the Results page and you selected some contact segments in step 1. Select this
option to export the selected contact segments only.
 Export first <X> contacts - to select a partial number of contact segments,
starting from the first.
 Random selection of <X> contacts - to produce a list of contact segments
selected at random.
6 If the check box is selected (default), the Saved Reports window is automatically
opened after you save the report. From the Saved Reports window, you can open
the report immediately upon its generation.
7 If you clear the check box, then after you save the report, the Create Report
window will be closed and you will return to the page from which you created the
report.
TIP You will be able to open the report from the Saved Reports window at any later
time by clicking the Saved Reports button in the Speech Analytics home page.
For information on managing saved reports, see “Managing Reports” on page 220.
8 Click OK.
Compare Two Result Sets Overview

Comparing Two Result Sets enables you to quickly create a report that compares data
from two different predefined Category or Saved Search result sets.
The purpose of this feature is to help you recognize the unique terms/phrases
associated with one Category or Saved Search and not the other. For example, when
you want to understand why a group of calls were successful and others were not, you
can create a Comparing Two Result Sets report to pinpoint what was unique about the
successful calls when compared to the unsuccessful calls.
The Comparing Two Result Sets report is displayed as a table in Excel. The table shows
the specific data field differences between the two selected result sets. For more
information, see “Comparing Two Result Sets Report Analysis” on page 217.
Typically the Comparing Two Result Sets report is created in up to 20 minutes. For more
information about how to create a Compare Two Results Sets report, see “Comparing
Two Result Sets” on page 216.
NOTE Each of the result sets should include at least 750 contacts.

Comparing Two Result Sets

1 From the Speech Analytics home page, or from the Results page, click the Create
Report button.
The Create Report window opens.
2 Select the Compare Two Result Sets tab.
3 Type the name of the Comparing Two Result Sets report in the Report Name field.
4 Select the type of result sets you want to compare (Category or Saved Search).
5 Hover over and click the Filter, Terms and/or Context icons in the table, to display
the type of filter and terms associated with the specific Category or Saved Search.
6 From the Category or Saved Search table, select and click one of the two result sets
you want to compare.
7 Click the Add button for Result Set 1.
8 From the Category or Saved Search table, select one of the two result sets you want
to compare.
9 Click the Add button for Result Set 2.
If the check box is selected (default), the Saved Reports window is automatically
opened after you save the report. From the Saved Reports window, you can open
the report immediately upon its generation.
If you clear the check box, then after you save the report, the Create Report
window will be closed and you will be returned to the page from which you created
the report. You will be able to open the report from the Saved Reports window at any later
time by clicking the Saved Reports button in the Speech Analytics home page.
For more information, refer to “Managing Reports” on page 220.
11 Click the Create Report window to create the report.
The Saved Reports window is opened after a few seconds.

Comparing Two Result Sets Report Analysis

The Compare Two Result Sets report enables you to recognize the terms/phrases that
differentiate one result set (Category or Saved Search) over another result set, by
showing the ratio of a word/phrase compared to another word/phrase in both results
sets.
NOTE The Comparing Two Results Sets report will include the top 50 results ordered
according to ratio.
Any results for which the ratio is <1.5% will not be included in the report.
The final Comparing Two Result Sets report will include the following details:
Data Description
Instance Name Indicates the unique name of the instance.
Created Indicates the report creation date.
Query History For a Category set, Query History contains the name of the
category and the number of contacts that make up its result set.
For a Saved Search set, Query History contains all the searches
made (that is, terms, phrases, categories, etc.), in chronological
order. The number represents the result of each search.
Effective Filter Indicates the filters selected for the query.
Contacts (Count) Indicates the amount of contacts associated with the result set.
Contacts (Percentage) Indicates the percentage of Count value out of the entire user
contact visibility size.
Contacts (Overlap) Indicates the intersection size of each set.

The lower the Overlap the better, since a low overlap indicates a big
difference between the two result sets.
Query Result Metrics Indicates the metrics of the two result sets.
Differentiating Terms Contains two tables:

The first table represents the terms/phrases that are unique to the
first result set compared to the second result set.
The second table represents the terms/phrases that are unique to
the second result set compared to the first result set.
Both tables are sorted according to the percentage of occurences
ratio from highest to lowest.
Both Differentiating Terms tables are made up of the following
columns and rows:
- Unique Terms
- % Occurrences
- % Occurrences Ratio
- White line
Unique Terms Indicates the word/phrase that appears in both result sets.

Data Description
% Occurrences Indicates the percentage of times the word/phrase appears out of

the total amount of contacts in the specific result set.
% Occurrences Ratio Indicates how relevant one word/phrase is in a result set when
compared with the same word/phrase in the other result set. For
example, Applications has an % Occurences Ratio of 2.08. Budget
has an % Occurences Ratio of 2.06. For this reason, the word
Applications is more relevant than the word Budget when the Good
result set is compared to the Account result set (see the example
below).
The % Occurrences Ratio is calculated by dividing the first result set
by the second result set in the first table and the second result set
by the first result set in the second table.
White line Indicates that the word/phrase is either a synonym, variant or

subset of the word/phrase in the line above.
The following is an example of comparing two Category or Saved Search results:

For more information, see “Compare Two Result Sets Overview” on page 215.


Chapter 8 - Reports and Data Export Managing Reports
Managing Reports
The Saved Reports window contains a list of all the reports you have created. From
this window, you can:
 View report details in each of the “Saved Reports Columns” on page 220
 “Refreshing the Report Status” on page 222
 “Saving a Report” on page 222
 “Deleting a Report” on page 222
Saved Reports Columns

The Report Status column gives a brief description of the export task status, and the
number and percentage of files that have been exported. The icons in the left most
column display a graphical status for each export task:
Icon Name Description
New Task Indicates a new task has

been created but has not
yet entered the queue.
In Queue Indicates that this task

has not been processed
yet.
In Process Indicates the export of

this task is still in progress
and shows the percent of
data that has been
processed.
Completed Indicates that this export

task is complete and
available for use.
Failed Indicates that this export

task has failed.

The Report Name column displays a link bearing the name you assigned the report
when you created it. Click the link to open the report. For Excel files, a dialog box
appears asking if you wish to open or save the file. If the file is a PDF file, it opens in
Adobe Acrobat.
The Description column displays a detailed description of the report. For example,
filters set, start and end dates, ANI, number of contacts included, etc. Click the
Description button to open the Description dialog box containing relevant report
details. Click the Close button to close the Description dialog box.
The Creation Date column displays the report's creation date and time, from newest to
oldest.
The Type column displays the type of report you requested: Categories, Contacts or
Charts.
The Format column displays icons for either Excel or PDF formats.
The Size column displays the file size, in kilobytes (KB) or Megabytes (MB). Files
smaller than 1 MB in size are displayed in kilobytes, and files larger than 1 MB are
displayed in Megabytes.
The Save As column allows you to download the report.
NOTE The Save As button appears only for reports that are completed, and does not
appear for reports that are in queue, in process or failed.
The Delete column allows you to delete a report, regardless of the current status of the
task.

Refreshing the Report Status

The Saved Reports window is opened before the report creation is complete.
The Saved Reports window contains a Report Status column that indicates (in
percentage), how much of the report has been created.
To refresh the Report Status:

1 From the Saved Reports window, click the Refresh button.
The report status is updated.
Saving a Report
1 From the Save As column, click the Save As button .
The Save As dialog box appears.
2 Select the path and file name and click OK.
The file is downloaded to the specified location.
Deleting a Report
1 From the Delete column, click the Delete button .
The Delete Task dialog box appears.
2 Click Yes to delete the export task.

Chapter 9
Ranking and Document

Relevancy
 Retrieving Information: Recall Vs. Relevancy, page 224

 Single and Multiple-Term Search Queries, page 225
 IDF Impact in Categories, page 228
Chapter 9 - Ranking and Document Relevancy Retrieving Information: Recall Vs. Relevancy
Retrieving Information: Recall Vs. Relevancy

The mechanism for retrieving information in Speech Analytics is designed to suit your
specific business needs.
Depending on your search entry, the mechanism adapts the way it calculates and
presents the search results:
 Search Results Based on Recall: In the recall search mechanism, search results
include all contacts that make the rank threshold in terms of their overall score,
regardless of the search term's frequency in the index.
In this case, a contact’s rank is determined by the number of instances of the
search term within the contact, and the contact’s individual score. No frequency-
based weighting is applied to the contact’s ranking.
This search mechanism is used for single term queries (such as "late charge"),
because when searching for a single term, the importance of the search is on the
recall of the term. Weighting related to a term's frequency is done between terms,
and therefore it is only relevant for search entries containing more than one term.
 Search Results Based on Relevancy: In searches based on the relevancy of the
term, the system bases its search logic on the industry-standard weight used widely
in information retrieval called Inverse Document Frequency (IDF). The IDF weight is
a statistical measure used to evaluate how important a term is in the index, which is
based on the term's frequency in the index. IDF diminishes the weight of terms that
occur very frequently, and increases the weight of terms that occur rarely. IDF is
used only in multiple-term entries (such as credit debt payment), as the
importance of the search is on the relevancy or power of the term within the index.
For multiple-term search entries, a contact's rank is determined by the number of
detected terms, their individual scores, and on the frequency of the term in the
index. If a term appears very frequently in the index, the IDF will be low. In this
case (because the IDF is low), the highest ranked contacts will be the ones that
contain many instances of that term, which then increases the overall score of the
contact.
The following table summarizes how each search mechanism calculates the search
results for each type of search query:
Search Mechanism Search Query Type Definition
Recall Single-term queries The system returns all contacts that make
the rank threshold in terms of their overall
score, regardless of the search term’s
frequency in the index (without IDF).
Relevancy Multiple-term queries The system uses IDF to determine the

relevancy and power of the term within the
index, diminishing the weight of terms that
occur very frequently, and increasing the
weight of terms that occur rarely.
For detailed information and use cases on the different types of search queries, see
“Single and Multiple-Term Search Queries” on page 225.

Chapter 9 - Ranking and Document Relevancy Single and Multiple-Term Search Queries
NOTE All category searches use the relevancy mechanism when calculating search
results (IDF). Therefore, for both single-term and multiple-term queries when
searching within categories, the system calculates the search results using IDF.
Single and Multiple-Term Search Queries

As described in “Retrieving Information: Recall Vs. Relevancy” on page 224, the system
retrieves information and calculates search results based on the type of search entry
you entered—a single term or phrase query, or a multiple term or phrase query.
The following table describes the differences between the two types of searches, the
operators and examples for each type of search:
Search Entry Definition Examples
Single-term Query  One word or phrase  Manager

 One word or phrase with  “late charge”
position-oriented  (call NEAR back) NOTIN
operators (such as NEAR, “call back today”
START, END, NOTIN)  [END:60]C: “thank you”
 [START](how NEAR can
NEAR help)
 [START:50]close NEAR
account
Multiple-term Query  More than one word or  angry annoyed

phrase  angry AND annoyed
 More than one word or  late -charge
phrase with AND, OR  late +charge
(indicated by a space) ,
NOT, minus (-) and plus
(+) operators
NOTE To indicate a single phrase query (including more than one word), use quotations
around the phrase (such as “late charge”). The system processes this as a single
phrase and calculates the search results accordingly (without the IDF weight).
To search for one word or another word (which is a multiple-term query), do not
use quotations around the phrase (such as credit debt). The system processes
this as a multiple-term query, and calculates the search results accordingly (with
the IDF weight).

For detailed examples of entering a single-term and multiple-term queries, see the
following use cases:
 “Single Term Query Default Use Case” on page 226: Enter a single-term query, with
the search results including all contacts that make the rank threshold in terms of
their overall score, regardless of the search term's frequency
 “Single Term Query Using Find Within Use Case” on page 227: Refine a single-term
query search by using the Find Within option, which uses the IDF weighting to
rank the search results, since the system evaluates it now as a multiple-term query
 “Single Term Query Using IDF Use Case” on page 227: Force IDF for single-term
queries
 “Multiple Term Query Use Case” on page 227: Enter a multiple-term query, with the
search results using the IDF weighting to find the relevant terms in the index
Single Term Query Default Use Case

The following describes a use case for entering a single-term search query:
1 In the Search field, enter the single-term or phrase with the relevant operators (as
described in “Single and Multiple-Term Search Queries” on page 225).
For example: “late charge”
2 By default, the system performs a pure, raw search of the term and returns all
contacts that make the rank threshold in terms of their overall score, regardless of
the frequency of the term. To indicate this search mechanism, an asterisk (*)
appears next to the number of search results.
For example: “late charge” (200)*
You can also view a tooltip that states that the search results were calculated
without using the IDF weight by placing your cursor over the term with the search
results.
These search results indicate that the system executed a search for all contacts
containing the term “late charge” and found 200 contacts that made the rank
threshold in terms of the overall score. In addition, a tooltip appears, explaining
that the search was performed without using the IDF weighting.

Single Term Query Using Find Within Use Case

The following describes a use case for refining a single-term query by using the Find
Within option. If you use this option, the system will use the IDF weighting to rank the
search results, because the system evaluates it now as a multiple-term query:
1 Enter your search entry (for example: “late charge”), and click Find.
2 As the search entry is a single term, the system searches for all contacts containing
the term within the rank threshold and will return the following search results:
“late charge” (200)*
These search results indicate that the system searched for all contacts that
contained “late charge” that made the rank threshold and found 200 contacts.
3 To refine your original search entry (“late charge”), enter another search entry
(for example: help), and click Find Within Results.
4 The system now treats the entry as a multiple-term query (“late charge” AND
help), and will return the following search results:
“late charge” (200)* > help (100)
These search results indicate that the system executed a new search for “late
charge” AND help and used the IDF calculation to find 100 relevant contacts with
the term.
Single Term Query Using IDF Use Case

The following describes a use case for forcing the IDF weighting for single-term
queries:
1 To override the default for single-term searches and to force the IDF weighting for
these types of searches, use the keyword: [IDF] before the search entry.
For example: [IDF] “late charge”
2 In this case, the system will perform the search using the IDF weighting, as in
multiple-term search queries (see “Multiple Term Query Use Case” on page 227),
and will present the results accordingly:
For example: [IDF] “late charge” (50)
These search results indicate that the system executed a search for “late charge”
using the IDF calculation and found 50 relevant contacts containing this term.
Multiple Term Query Use Case

The following describes a use case for entering a multiple-term search query:
1 In the Search field, enter the multiple-term or phrase with the relevant operators
(as described in “Single and Multiple-Term Search Queries” on page 225).
For example: angry annoyed

Chapter 9 - Ranking and Document Relevancy IDF Impact in Categories
2 The system uses the IDF weighting to calculate the search results, as the
importance of the search is on the relevancy or power of the term within the corpus.
For example: angry annoyed (100)
In this case, the system translates this entry into: angry OR annoyed, and
therefore, calculates the search results using IDF. The system found 100 relevant
contacts in the corpus.
IDF Impact in Categories

When analyzing categories, the system uses the IDF weight to calculate and prioritize
the category results.
Each contact is ranked according to its relevancy to the category, indicated by a defined
number of stars (5 being the highest rank). The higher the rank, the more relevant that
contact is for the category. Depending on the minimum rank setting, contact segments
below the rank threshold (2-star rank, for example) are excluded from the category
results.
A contact's ranking in a category, however, is not fixed. Approximately once an hour, the
system re-evaluates categories as new contacts are continuously being added to the
index. The addition of new contacts to the index can influence the ranking of the calls for
a specific category. This can result in different ranking for the same contact, depending
on when the category search was performed. In addition, it can also slightly change the
number of contacts in the result set.
Therefore, theoretically, it may happen that a contact that currently belongs to a specific
category will no longer belong to that same category at a later time. This could happen
if any of the category term frequencies have changed drastically due to the indexing of
new contacts, causing the term's IDF to change, which then changes the contact's
overall rank.
However, because the IDF calculation is only based on a timeframe of 30 days
backwards from each individual contact’s date, term frequencies are expected to be
stable; this is because a change in a term's frequency due to new contacts would not
affect the scoring of contacts older than 30 days. For the same reason, category
statistics are expected to be stable as well.

Chapter 10
Troubleshooting
This chapter outlines the following troubleshooting topics:

 Frequently Asked Questions, page 230
 Category Troubleshooting, page 230
 Charts Troubleshooting, page 233
 Error Messages, page 234
 Filter Troubleshooting, page 238
 General Troubleshooting, page 240
 Playback Troubleshooting, page 242
 Speaker Separation Troubleshooting, page 243
 Searches Troubleshooting, page 246
 Trends Troubleshooting, page 249
Chapter 10 - Troubleshooting Frequently Asked Questions
Frequently Asked Questions

The following list provides troubleshooting tips for FAQ’s you may encounter:
 When do New Transcription Rules Start to Work?, page 230
 What Happens to a Transcription Rule if a Group that is used in it has been Deleted
from the System?, page 230
 Can I Define a Rule for a Root-Level Group?, page 230
When do New Transcription Rules Start to Work?

While defining a new rule, you specify the date on which you want to enable the rule,
and your rule will be working from the specified day onward. If you specify the today's
date, the rule will be activated within an hour upon being created.
What Happens to a Transcription Rule if a Group that is used in

it has been Deleted from the System?
When a group that participates in a rule is deleted, the rule continues to work, provided
it contains any other existing groups. If all groups participating in the rule are deleted,
the rule will still exist in the system but will not work, since there will be no situations in
which this rule will apply.
Can I Define a Rule for a Root-Level Group?

Yes. You can define transcription rules for any level groups - both leaf groups containing
agents and higher groups containing other groups.
Category Troubleshooting
The following list provides troubleshooting tips for Category problems you may
encounter:
 What is the Difference between the Graphical and Classic Views?, page 231
 What is the Difference between Using Categories and Filters?, page 231
 The Difference between the Autocompletion Suggestions and the More Suggestions,
page 231
 Importing / Exporting Category Filters, page 232

Chapter 10 - Troubleshooting Category Troubleshooting
What is the Difference between the Graphical and Classic

Views?
The Graphical view and the Classic view are two different formats of defining category
terms in Speech Analytics.
For more in-depth information see “Category Term Definition: Classic vs. Graphical
What is the Difference between Using Categories and Filters?

The Speech Analytics application can display relevant business issues when searching
with filters or when using categories.
Searching using filters or drilling down into categories can help produce different sets of
calls, reflecting a variety of call issues as well as different facets of your business
operation.
Filters
Filters enable you to retrieve contacts of a specific type, based on metadata, i.e.,
contacts that were recorded on a specific date, that were handled by a specific agent, or
even contacts that belong to a specific category.
Filters can help you focus your category on a predefined set of calls relating to your
business issue, based on call information already available in the system.
For more information on Filters see “Defining Search Filters” on page 93.
Category
A Category is a type of content with which you can tag calls based on a predefined set of
rules. Calls are allocated to categories based on filters, terms and rules. Categories
themselves may be used as any other metadata, enabling queries and trend analysis.
A category definition may include a list of specific terms or phrases as well as filters.
Categories help retrieve contacts pertaining to a specific business issue and can identify
different call issues, i.e., Non Resolved Calls, Emotions, Complaints, as well as content
issues, i.e., Retention Issues, Competition or New Customer Needs.
By using Filters in categories, we can focus on specific call types, such as agent groups,
specific time frame.
For more information on Categories see “Creating Speech Analytics Categories” on
page 170.
The Difference between the Autocompletion Suggestions and

the More Suggestions
The Speech Analytics application offers Autocompletion when searching. As you type a
word, the Speech Analytics application starts searching and automatically displays a list
of the possible word's completions.
More Suggestions is a link at the right-hand side of the search.
Autocompletion is a list of words that are semantically related by word stems. It ensures
correct word spelling and decreases the amount of time spent typing words. In addition,

Chapter 10 - Troubleshooting Category Troubleshooting
the Autocompletion list, which is compiled from words that were detected in transcribed
calls, gives you suggestions regarding other words that Speech Analytics identified as
possibly related to the word you are typing, and provides valuable information about
these words' usage.
The More Suggestions drop-down list provides a context based suggestion list for your
text queries – during both keyword searches and category term definition. The
suggestions give you ideas on other related words to search for and can help you adjust
your query or come up with a better alternative for your word.
For more in-depth information about the Autocompletion list and the More Suggestions
list see “Guided Search” on page 77.
Importing / Exporting Category Filters

When importing/exporting a category with a defined filter from one system to another,
the following category filter parameters data will not be available for the imported/
exported category:
 Agent Data > Agent
 Agent Data > Group
 Advanced Data > Switch
Values for the following Category Filter parameters are system dependent:
 Advanced Data > Custom Data
 Advanced Data > Conditional Custom Data
 Date Range > Time of Day
If both systems have the same value definitions for these parameters, the category filter
parameters data can be imported/exported. If the two systems do not have the same
value definitions for these parameters, unexpected category results appear.
When importing/exporting a category that contains a filter of type Context > Speech
Analytics Categories, the source instance number and the destination instance number
must be the same.
Categories that are imported with invalid filter definitions, may contain unexpected
results. To produce accurate results from such categories, you must configure the
category filter manually.

Chapter 10 - Troubleshooting Charts Troubleshooting
Charts Troubleshooting
The following list provides troubleshooting tips for Chart problems you may encounter:
 The Average Line Value in the Charts is Different from the Average Value, page 233
 The Chart Does Not Appear, page 234
The Average Line Value in the Charts is Different from the

Average Value
As shown in the following image, the average line value of the following charts is
different from the average value in the top right corner of each chart.
 Duration
 Number of Holds
 Number of Transfers
 Total Hold Time
 Agent Talk-Time Percentage
 Customer Talk-Time Percentage
 Silence-Time Percentage
 Talk-Over-Time Percentage
The line that appears across the charts represents the average number of contact
segments per group.

Chapter 10 - Troubleshooting Error Messages
The average value at the top of the charts represents the average of the examined
parameter, for example the average duration.
It is important to note that the Speaker Separation charts only include segments
recorded with the Speaker Separation feature. Therefore, an asterisk appears next to
the calculated average.
For additional information see “Using Charts” on page 127.
The Chart Does Not Appear

When trying to view a chart in the Speech Analytics application home page or the
Results page, an Internet Explorer time-out event may occur, whereby the chart does
not appear.
To view the chart, right-click the place where the chart should be and select Show
Picture from the menu.
Error Messages
The following list provides troubleshooting tips for problems you may encounter:
 The Current Speech Analytics Session has Expired. Please reopen the Speech
Analytics Application, page 234
 The following message appears: Due to network problems this contact's
transcription could not be retrieved., page 235
 Action failed around 3 am, page 235
 Media Components Cannot be Accessed for Playback, page 235
 The Underlying Connection was Closed. Unable to Connect to the Speech Analytics
Server., page 237
 Login Failed. Your Account is Probably not Defined Properly. Contact your System
Administrator for more Information., page 237
 Adobe Flash Player is Required to View Trend Charts. You can Download the Player
from the Adobe Downloads Site., page 237
 The Speech Analytics Application cannot be opened. Connection to the Server or
one of its Components is Down. Please try to Open Speech Analytics at a Later
Time. If the Problem Persists, Contact your System Administrator., page 238
The Current Speech Analytics Session has Expired. Please

reopen the Speech Analytics Application
The Speech Analytics Portal cannot be idle for more than a configurable amount of
minutes. If Speech Analytics is idle for more than the configurable amount of minutes,
the user is redirected to a login page. Once the user’s credentials (password and
username) are entered, the page is loaded again.

Due to network problems this contact's transcription could not

be retrieved.
This message appears when a disconnection occurs between the Speech Analytics
Application Server and the user's desktop. In this situation, the audio and screens may
still be played back.
Action failed around 3 am

If you log into the Portal before 3 am and continue to work, you may encounter some
problems that cause your action to fail. (For example, you may fail to submit an
evaluation or fail to open a contact from the search results list etc.) The problem may be
caused by your system's maintenance activities, which occur by default at 3 am. If
required, your system administrator can change the time these maintenance activities
are performed.
You will receive a message that describes the error and what corrective actions must be
performed. Messages for example: "User profile or system definitions have been
modified. Please log out and log in again for updated changes to take effect. If the
problem persists, contact the System Administrator" or "Internal error. Notify the
System Administrator to contact the help desk" or "Your Web Portal session has expired.
Please close all open browser windows and log in to your system again".
1 In the Internet Information Services (IIS), select the relevant application pool.
2 In the Recycle tab, select the default time and click Remove.
3 Enter the preferred time and click Add.
4 Click OK.
Media Components Cannot be Accessed for Playback

When media components cannot be accessed for playback the following error message
appears:
"At least one media component could not be accessed for playback (audio and/or
screens). Click OK to close this message, and then click the play icon. The existing
media components will be played back." message appears when you:
 Try to playback from the search results page or folder list.
 Try to playback a contact in the workspace page.
This message means that your workstation failed to access one of the contacts
components (audio or screens). If the contact indeed has both media components, you
can close the message and click Play. If one of the components was successfully
accessed, it will be played back.
1 Check whether the file was downloaded to the workstation’s Temporary Internet
Files folder as follows:
a. Open Internet Explorer.
b. From the Tools menu, select Internet Options. The Internet Options dialog
box appears.

c. In the Temporary Internet Files group box, click Delete Files. The Delete
Files dialog box appears.
d. Select the Delete all offline content check box and click OK. You return to the
Internet Options dialog box.
e. Click Settings. The Settings dialog box appears.
f. Click View Files. Your Temporary Internet Files folder opens in Windows
Explorer.
g. Repeat the action that caused the message to appear (for example, open the
contact again in the Results page).
h. Search your Temporary Internet Files folder (opened in step f above) and all
sub-folders for the downloaded contact. Search for the *.wav or *.vox.wav (for
audio files) or *.avi (for audio+screen files) extension. Your system files are
named in the following format: XX_YY_start-time. XX is the Acquisition Module
and YY is the channel.
If you found the file you were trying to playback, proceed to steps 2-3. If the is
not found, proceed to step 4.
2 If the file is found in the folder, it means the file was successfully downloaded to the
server. Verify the following IT issues:
 Make sure your recording platform's DNS (host) name is configured in your DNS
Server or local hosts file. After you fix this, try to repeat the action that caused
the error (for example, return to the search results list and select the file again).
 If your system includes recording on physically remote sites, where the storage
servers are in a different domain, verify there is a trust relationship between the
workstation's domain and the domain you are trying to access for playback.
3 Now you need to determine why the file cannot be played back. Double-click the file
to play it back using the Media Player.
 If you succeed, this indicates a problem in playing back from your Speech
Analytics platform, likely caused by the current Internet settings. Compare the
proxy server settings to a workstation that can playback. Your Speech Analytics
platform and the audio and screen storage servers must be in the proxy server's
exception list.
 If you cannot playback the file, try to playback a downloaded contact using
Media Player.
If you succeed, this indicates a problem in your recording platform playback. Try
to uninstall the Multimedia Support Package, and then access the Speech
Analytics application again to begin the automatic download of the package.
If you fail, this indicates a general playback problem on this workstation. Contact
your organization's technical support for assistance.
4 If the problem persists, additional file access troubleshooting is required. Contact
technical support.

The Underlying Connection was Closed. Unable to Connect to

the Speech Analytics Server.
This error message may appear for the following:
 Disconnection between the Speech Analytics WCF Host Service in the Speech
Analytics Application server and the Speech Categorization service.
 The Speech Analytics WCF Host Service is down.
 The Speech Analytics Application server is down, therefore thre is no access from
the server hosting the server role to the Speech Analytics WCF Host Service.
The following solves each of the reasons listed above:
1 From System Tools’ Run tab, open the Watchdog Control Panel.
2 In the Watchdog Control Panel stop and then restart the Speech Categorization
service.
3 In the Watchdog Control Panel, select Stop and then Restart the WCF service.
4 If problem persists, contact your technical support.
Login Failed. Your Account is Probably not Defined Properly.

Contact your System Administrator for more Information.
Working with Speech Analytics requires that you have permission to do so. If you were
not give permissions (for example, rights) for Speech Analytics you will not be able to
log into it.
To work with Speech Analytics, give the user the relevant permissions in Assignment
Manager, refer to the Assignment Manager Online Help for more details.
Adobe Flash Player is Required to View Trend Charts. You can

Download the Player from the Adobe Downloads Site.
When attempting to view a Trend chart the following error message appears:
Error Message: Adobe Flash Player is Required to View Trend Charts. You can Download
the Player from the Adobe Downloads Site.Please Download and Install Latest Flash
Version From the Following Location: http://www.adobe.com/go/getflashplayer.
Adobe Flash Player is required to view Trend charts.
Verify that Adobe Flash Player Version 9 and above has been installed.

Chapter 10 - Troubleshooting Filter Troubleshooting
The Speech Analytics Application cannot be opened.

Connection to the Server or one of its Components is Down.
Please try to Open Speech Analytics at a Later Time. If the
Problem Persists, Contact your System Administrator.
Contact the your System Administrator and refer to the System Administration Online
Help for more details.
Filter Troubleshooting
The following list provides troubleshooting tips for Filter problems you may encounter:
 Cannot Find a Contact Group, page 238
 Why Can’t I See All Groups?, page 238
 TellMeWhy Analysis is Not Providing Results, page 238
 Groups and Agents, No Search Results are Returned, page 239
 Why are Expected Contacts Missing, When Performing a New Search?, page 239
Cannot Find a Contact Group

If you are trying to access a transcribed contact in the Speech Analytics application and
are using the group as a filter, note the following behavior: in the Speech Analytics
application, a contact is associated to a group according to the agent's group association
on the day the contact was recorded. If the agent has since been assigned to a new
group, searching by the new group will not return this contact. Instead, search for the
contact by entering the agent's name only, without selecting a group.
For example, if Sara belonged to Group A on Monday, but was moved to Group B on
Tuesday, if you perform a search on Thursday for her calls recorded on Monday, search
by Group A. Alternatively, just search by Sara's name, and do not use the group as a
filter.
Why Can’t I See All Groups?

In the Speech Analytics application, you can only view contacts made by agents that
have either been directly assigned to your group(s), or that are assigned to subgroups
of your groups. Therefore, when setting filters, you will not be able to see the entire
organization, unless you were assigned to a group at the top of the organizational
hierarchy, and all search results will include contacts belonging to these agents only.
Group assignment is performed in the Assignment Manager.

Chapter 10 - Troubleshooting Filter Troubleshooting
TellMeWhy Analysis is Not Providing Results

TellMeWhy™ analysis helps identify underlying drivers of performance, and determines
opportunities for improving efficiency and quality in your business. For additional
information about TellMeWhy™ analysis see “Viewing TellMeWhy Analysis” on page 160.
TellMeWhy™ analysis is based on the content of the calls and the correlation between
the contacts in the search results. If content correlation is weak or ill-structured,
TellMeWhy™ analysis cannot offer suggestions (for example, the focus of TellMeWhy™
analysis is call characteristics). If the characteristics of the calls are not significantly
related, suggestions of a correlation cannot be offered. When these conditions are
present and the TellMeWhy™ analysis is requested, the following message appears:
In addition, the amount of contacts listed in your results set may also influence the
TellMeWhy™ analysis ability to offer suggestions. In a small result set, there may not be
enough contacts to find correlations. For that reason, the TellMeWhy™ analysis will not
be able to provide results. In this case, you may wish to modify your search query to
include a larger, more diverse set of calls and re-apply the TellMeWhy™ analysis.
Groups and Agents, No Search Results are Returned

In such a scenario, the Speech Analytics application first checks whether the selected
agents belong to the selected group before searching for the contacts. If the selected
agents do not belong to the selected group the search is aborted and therefore no
results are returned.
For this reason, every time you set a filter for a category or a search, verify that the
selected agents are part of the selected group. Do not select an agent that is not part of
the selected group.
For more in-depth information, see “Filter List Parameters” on page 95.
Why are Expected Contacts Missing, When Performing a New

Search?
When performing a new search, you might be seeing a subset of the Expected results
list. For example, not all of the contact segments you expected will appear in the list.
The reason you are not seeing the full result list for your new search is that Filters for
your previous search are still enabled. Filters are not automatically cleared every time
you create a new search.
If Filters are currently set, the bar above the search results will contain the following:
 Filters are currently set
 Filter icon .
To view the filters that are currently set, place your mouse over the Filter icon above the
results list.
For new searches that do not require previous filters, clear the filters by clicking the

Chapter 10 - Troubleshooting General Troubleshooting
For additional information see “Defining Search Filters” on page 93 (specifically the
image at the bottom of the page).
General Troubleshooting
The following list provides troubleshooting tips for problems you may encounter:
 What is the Difference Between the Home Page and the Landing Page?, page 240
 User and Password Prompts Appear, page 240
 Troubleshooting Popups Appear with a Status Bar in Trusted Sites, page 241
 Why Do Tool Bar Options Periodically Appear?, page 241
What is the Difference Between the Home Page and the

Landing Page?
The Trend Analysis tab is displayed when you first open the Speech Analytics
application. For this reason, it is referred to as the Landing page.
When the Home page button is clicked, the Contact Analysis page appears.
The Trend Analysis tab is the Landing page, because it presents the user with up to date
information about the system. For example, the Trend Analysis tab will show the user
which word’s usage changed significantly.
The Contact Analysis page is the Home page, because its features and contents are
most frequently used for contact search and analysis.
For more in-depth information about these pages refer to:
User and Password Prompts Appear

When using the Portal applications, a dialog box appears, requiring you to enter your
user name and password. Select the instance when this occurs:
During Playback
1 If your system setup does not include trust between the Portal domain and your
own domain, you will always receive this prompt. Enter your login details to
continue.
2 For IT administrators: the Portal is trying to access audio and screens files on its
storage servers using the http protocol. Storage servers include the servers where
the Voice or VoIP Acquisition Modules, Audio Transfer and Storage Module and
Screen Acquisition Module are installed.
a. Ensure the storage servers are defined as an Intranet sites. This is achieved by
one of the following setups:

Chapter 10 - Troubleshooting General Troubleshooting
 Add the storage servers to the list of local Intranet addresses. On each
workstation, from Internet Options, click the Security tab. Select Local
Intranet and click Sites. Click Advanced, and add the storage servers' IP
addresses or FQDNs to the zone. This must match the way your system has
been configured (IP vs. FQDN). Contact technical support to check your
system's configuration.
 If you use a proxy, add the storage servers' IP address or FQDN to the
Proxy's exception list. On each workstation, from InternetOptions, click the
Connections tab. Click LAN Settings. Ensure the Bypass proxy server
for local addresses check box is selected, and click Advanced. Add the
storage servers' IP addresses or FQDNs to the Exceptions list. This must
match the way your system has been configured (IP vs. FQDN). Contact
technical support to check your system's configuration.
b. Ensure your local Intranet security settings are configured with the Automatic
logon option. On each workstation, from Internet Options, click the Security tab.
Select Local Intranet and click Custom Level. Scroll down to UserAuthentication,
Logon and select one of the Automatic Logon options.
3 Tips for Windows XP and Windows 2000: You can save your user name and
password for automatic logon. In Windows XP this eliminates the prompt, and in
Windows 2000 it remembers the details and you only need to click OK. The next
time the prompt appears, select the Remember my password check box.
When Returning from the Speech Analytics Application

Contact technical support to verify the system configuration. The Portal requires all the
servers involved in web browsing be configured in the same way with regards to naming
(DNS, FQDN or IP addresses). You must browse to the server in the same way.
Troubleshooting Popups Appear with a Status Bar in Trusted

Sites
Popups open with a status bar in a trusted site. This occurs on Windows XP SP2. To
resolve this issue you must enable "Allow script initiated windows without size or
position constraints" in the trusted zone's security properties.
1 In the Internet Explorer, click on Tools and select Internet options.
2 Select the Security tab and click on the Custom Level button.
3 Scroll down to the Allow script initiated windows without size or position
constraints option and click on the Enable radio button.
4 Click OK.
5 Click Yes.
6 Click OK.

Chapter 10 - Troubleshooting Playback Troubleshooting
Why Do Tool Bar Options Periodically Appear?

The toolbar at the top of the Speech Analytics window displays buttons that allow you to
access functionality relevant to the tasks you are performing and to the information
displayed.
When you are working in the Trend Analysis tab or the Contact Analysis tab (in its Home
Page), the main toolbar is displayed. The main toolbar allows you to access the Category
Builder, saved searches, and filtering and reporting functionality of the application.
The buttons on the toolbar change when you are viewing the Results page. For more
information on the Results page toolbar, see “Understanding the Results Page” on
page 107.
Playback Troubleshooting
The following list provides troubleshooting tips for Playback problems you may
encounter:
 Why is it that I Can only Listen to One Segment When I Have a Metrics for the
Entire Contact?, page 242
 There is No Play Icon in the Results Page, page 242
 The Player is Not Resized with the Browser Window, page 243
 Why Do the Volume Settings of the Player Change?, page 243
 Why Do the Screens Appear and Disappear in the Player?, page 243
Why is it that I Can only Listen to One Segment When I Have a

Metrics for the Entire Contact?
I have query result metrics for the entire contact, but I am only listening to one
segment. For example, I cannot see which other segments are part of the contact.
During a contact, each time the customer is transferred, a new segment begins. The
Speech Analytics Application deals with contact segments that are analyzed for business
issues. When working with the Speech Analytics Application, you focus on business
needs that are analyzed per segment.
In the query result metrics, you are provided information per the specific segment as
well as certain metrics regarding the whole contact, such as Average number of holds,
Average number of transfers, Average duration, Average hold time, etc.
To playback the entire contact, access the Portal and use the Quality Monitoring or the
Customer Xperience application to do so.
There is No Play Icon in the Results Page

If you have been assigned the "playback filtered on contacts only" right in the
Assignment Manager, you will only be able to playback contacts that you have been
granted access to via the contact filters that are defined in the Assignment Manager. The

Chapter 10 - Troubleshooting Playback Troubleshooting
Speech Analytics application enables you to search for all transcribed contacts but you
can only playback contacts that are defined as part of your filter.
The Player is Not Resized with the Browser Window

If you expand the browser window at the time when the segment is being played, the
size of the Speech Analytics Player does not change proportionally to the full window
size, but remains the same.
To have the player size proportional to the browser window size, repeat your search and
then expand the browser window (or resize it as required) before starting the playback.
See “Searching in Speech Analytics” on page 66.
Why Do the Volume Settings of the Player Change?

When playing back a segment in the Speech Analytics Player, you can adjust the volume
of the playback by using the Volume control of the player. This setting is maintained for
the playback of all segments from the same search results.
However, when you perform a new search, and then play any of the retrieved segments,
the Speech Analytics Player is restored to its default volume.
Why Do the Screens Appear and Disappear in the Player?

By default, when playing back a segment with recorded screens, the Speech Analytics
Player does not display the screens. If you wish to view the screens, you can display
them in a separate window by clicking the View Screens button .
If, after displaying screens for a segment playback, you continue playing segments from
the same search results, the segments will be played with screens.
However, when you perform a new search, the Speech Analytics Player is restored to its
default settings, and the screens are not automatically displayed during the playback.
For more information on the screen playback, see “Viewing Screens” on page 157.

Chapter 10 - Troubleshooting Speaker Separation Troubleshooting
Speaker Separation Troubleshooting

The following list provides troubleshooting tips for Speaker Separation problems you
may encounter:
 Why is it that I Can Only Listen to One Segment When I Have a Metrics for the
Entire Contact?, page 244
 Some of the Contacts in the Search Results do not Include Speaker Separation,
page 244
 Failing to Search for Agent or Customer Specific Terms, page 244
 The Average Line Value in the Charts is Different from the Average Value, page 245
 Number of Contacts Appearing in the Speaker Separation Chart is Different from the
Number of Retrieved Contacts, page 246
Why is it that I Can Only Listen to One Segment When I Have a

Metrics for the Entire Contact?
I have query result metrics for the entire contact, but I am only listening to one
segment. For example, I cannot see which other segments are part of the contact.
During a contact, each time the customer is transferred, a new segment begins. The
Speech Analytics Application deals with contact segments that are analyzed for business
issues. When working with the Speech Analytics Application, you focus on business
needs that are analyzed per segment.
In the query result metrics, you are provided information per the specific segment as
well as certain metrics regarding the whole contact, such as Average number of holds,
Average number of transfers, Average duration, Average hold time, etc.
To playback the entire contact, access the Portal and use the Quality Monitoring or the
Customer Xperience application to do so.
Some of the Contacts in the Search Results do not Include

Speaker Separation
When playing back certain contact segments, you may notice that some of them show
separate flows for the Agent and Customer and others do not. That is, although the
Speaker Separation feature is enabled in your system, some of the contact segments
are not displaying it.
Although Speaker Separation is enabled in your system, your system may be working in
mixed mode. For example, it contains segments recorded with Speaker Separation and
segments recorded without Speaker Separation.

This can occur when:

 Only some of the recording units have been configured with Speaker Separation and
therefore segments recorded by Non Speaker Separation configured acquisitions
units will not display Speaker Separation properties.
 Your system was updated from earlier versions of Speech Analytics where the
Speaker Separation feature was not supported and therefore the segments do not
display Speaker Separation properties.
Failing to Search for Agent or Customer Specific Terms

When searching for keywords specifically spoken by an Agent or a Customer, the text
queries that perform the search are built using the A: or C: operator respectively.
The results may not appear due to spacing issues when entering the operators. Ensure
there is no space between the semicolon (:) and the keyword(s) for which you are
searching. If there is a space after the semicolon the search will not work.
For example, A:help will work whereas, A: help will not.
The Average Line Value in the Charts is Different from the

Average Value
As shown in the following image, the average line value of the following charts is
different from the average value in the top right corner of each chart.
 Duration
 Number of Holds
 Number of Transfers
 Total Hold Time
 Agent Talk-Time Percentage
 Customer Talk-Time Percentage
 Silence-Time Percentage
 Talk-Over-Time Percentage

The line that appears across the charts represents the average number of contact
segments per group.
The average value at the top of the charts represents the average of the examined
parameter, for example the average duration.
It is important to note that the Speaker Separation charts only include segments
recorded with the Speaker Separation feature. Therefore, an asterisk appears next to
the calculated average.
For additional information see “Using Charts” on page 127.
Number of Contacts Appearing in the Speaker Separation

Chart is Different from the Number of Retrieved Contacts
This topic is relevant for charts that are specific to Speaker Separation.
With Speaker Separation enabled, your system may be working in a mixed mode. For
example, it contains segments recorded with Speaker Separation and segments
recorded without Speaker Separation. This can occur if some of the recording units have
been configured with Speaker Separation and others have not, or when your system
was updated from earlier versions of Speech Analytics.
In such systems, Speaker Separation data is calculated and displayed only for segments
recorded with Speaker Separation. For example, if both types of segments appear in the
search results, the Speaker Separation data metrics are calculated only on those
segments that were recorded with Speaker Separation.
Segments recorded without Speaker Separation, if such are included in the result set,
are omitted from the chart. As a result, the number of segments in such a chart is lower
than the number of the segments in the result set.
Example, lets say you have 1000 calls but only 800 contain Speaker Separation. If you
view the Agent Talk time chart (Speaker separation metric) in the Home Page the chart
will be calculated according to 800 contacts.

Chapter 10 - Troubleshooting Searches Troubleshooting
For additional information, see “Speaker Separation” on page 18.
Searches Troubleshooting
The following list provides troubleshooting tips for Search problems you may encounter:
 Why are Expected Contacts Missing, When Performing a New Search?, page 246
 Viewing More than 10,000 Search Results, page 247
 Search Results are Not Accurate, page 247
 Why Can't I Search by Keyword?, page 248
 No Contacts are Available, page 248
 TellMeWhy Analysis, page 159
 When Filtering by Groups and Agents, No Search Results Are Returned, page 248
 Difference Between the Autocompletion Suggestions and the More Suggestions,
page 249
Why are Expected Contacts Missing, When Performing a New

Search?
When performing a new search, you might be seeing a subset of the Expected results
list. For example, not all of the contact segments you expected will appear in the list.
The reason you are not seeing the full result list for your new search is that Filters for
your previous search are still enabled. Filters are not automatically cleared every time
you create a new search.
If Filters are currently set, the bar above the search results will contain the following:
 Filters are currently set
 Filter icon .
To view the filters that are currently set, place your mouse over Filter icon above the
results list.
For new searches that do not require previous filters, clear the filters by clicking the
For additional information see “Defining Search Filters” on page 93 (specifically the
image at the bottom of the page).
Viewing More than 10,000 Search Results

When viewing the search results, note that for each search, the results are ranked and
then presented in descending relevancy of the assigned rank. To learn more about
ranks, see “Understanding the Results List” on page 109. The total number of results
and the percentage of calls these represent are always displayed at the top of the
results page, but only the top 10,000 contacts are sorted and presented. The total

number and percentage are also available in the Speech Analytics reports. For more
information on reports, see “” on page 212.
If a review of more than 10,000 individual calls is required on a particular search or
topic, the search can easily be split into several searches. For example, you can perform
the search by smaller time units. If the search was intended for the last three weeks,
you can split it to include only one week at a time. For searches that may generate even
more results, you can split the results by day or even by time of day, duration or any
mutually exclusive meta-data.
Search Results are Not Accurate

If some information which you were looking for was updated in the last hour, the
updates may not be available yet from the Speech Analytics application. That's why the
search results are not accurate.
The availability of this information in the Speech Analytics application depends on the
type of the search.
In searches that use the <All> options and do not specify any particular parameters,
the updated information becomes available within an hour upon the change.
For example, if an agent was deleted in the last hour, and you perform a search for all
groups immediately after the deletion, the deleted agent will still appear in the Portal.
Similarly, if an agent was transferred to another group, the search results will still show
the agent in the previous group.
Yet, searches in which you specify the parameters to retrieve and do not use <Any>
options are always up-to-date and bring the most update information.
1 While performing a search, do not use the <All> options, but specify the
parameters to search for.
2 If you want to use the <All> option, perform the search in about one hour after the
information you are looking for was modified.
Why Can't I Search by Keyword?

The ability to perform a search according to keywords or phrases is controlled by a right
that must be assigned to you in the Assignment Manager. If you do not have the Search
by Keyword right, you will not be able to perform keyword searches, nor will you be able
to use the TellMeWhy™ analysis.
No Contacts are Available

When you open Speech Analytics, the number of transcribed contacts currently available
is indicated. If no contacts are available:
 Verify that you have opened the Speech Analytics Management Tool, and that an
index has been activated, before opening the Speech Analytics application.
 Ensure that the active index is configured properly in the Speech Analytics
Management Tool.

TellMeWhy Analysis Is Not Providing Results

TellMeWhy™ analysis helps identify underlying drivers of performance, and determines
opportunities for improving efficiency and quality in your business. For additional
information about TellMeWhy™ analysis see “Viewing TellMeWhy Analysis” on page 160.
TellMeWhy™ analysis is based on the content of the calls and the correlation between
the contacts in the search results. If content correlation is weak or ill-structured,
TellMeWhy™ analysis cannot offer suggestions, i.e., the focus of TellMeWhy™ analysis is
call characteristics. If the characteristics of the calls are not significantly related,
suggestions of a correlation cannot be offered. When these conditions are present and
the TellMeWhy™ analysis is requested, the following message appears:
In addition, the amount of contacts listed in your results set may also influence the
TellMeWhy™ analysis ability to offer suggestions. In a small result set, there may not be
enough contacts to find correlations. For that reason, the TellMeWhy™ analysis will not
be able to provide results. In this case, you may wish to modify your search query to
include a larger, more diverse set of calls and re-apply the TellMeWhy™ analysis.
When Filtering by Groups and Agents, No Search Results Are

Returned
In such a scenario, the Speech Analytics application first checks whether the selected
agents belong to the selected group before searching for the contacts. If the selected
agents do not belong to the selected group the search is aborted and therefore no
results are returned.
For this reason, every time you set a filter for a category a search, verify that the
selected agents are part of the selected group. Do not select an agent that is not part of
the selected group.
For more in-depth information, see “Filter List Parameters” on page 95.
Difference Between the Autocompletion Suggestions and the

More Suggestions
The Speech Analytics application offers Autocompletion when searching. As you type a
word, the Speech Analytics application starts searching and automatically displays a list
of the possible word's completions.
More Suggestions is a link at the right-hand side of the search.
Autocompletion is a list of words that are semantically related by word stems. It ensures
correct word spelling and decreases the amount of time spent typing words. In addition,
the Autocompletion list, which is compiled from words that were detected in transcribed
calls, gives you suggestions regarding other words that Speech Analytics identified as
possibly related to the word you are typing, and provides valuable information about
these words' usage.
The More Suggestions drop-down list provides a context based suggestion list for your
text queries – during both keyword searches and category term definition. The

Chapter 10 - Troubleshooting Trends Troubleshooting
suggestions give you ideas on other related words to search for and can help you adjust
your query or come up with a better alternative for your word.
For more in-depth information about the Autocompletion list and the More Suggestions
list see “Guided Search” on page 77.
Trends Troubleshooting
The following list provides troubleshooting tips for Trends problems you may encounter:
 What is the Difference Between the Home Page and the Landing Page? on page 249
 Trend Analysis Results are Not Available on page 250
What is the Difference Between the Home Page and the

Landing Page?
The Trend Analysis tab is displayed when you first open the Speech Analytics
application. For this reason, it is referred to as the Landing page.
When the Home page button is clicked, the Contact Analysis page appears.
The Trend Analysis tab is the Landing page, because it presents the user with up to date
information about the system. For example, the Trend Analysis tab will show the user
which word’s usage changed significantly.
The Contact Analysis page is the Home page, because its features and contents are
most frequently used for contact search and analysis.
For more in-depth information about these pages refer to:
Trend Analysis Results are Not Available

Trend analysis returns results only if the number of contact segments for the selected
time frame is sufficient and only during time periods in which trend history is available.
The minimal data thresholds are:
 Daily analysis - 3 weeks' worth of data.
 Weekly analysis – 4 weeks' worth of data.
 Long-term analysis (from month to year) - a worth of data of the selected time
period. For example, for monthly trend analysis, a month's worth of data is
required.
When calculating trend statistics, the Speech Analytics solution excludes from its
calculations days with exceptionally low call volumes. These are considered days with
less than 700 calls or the days in which the number of calls is significantly lower than
the median number of calls from the date when the tracking started.

Chapter 10 - Troubleshooting Trends Troubleshooting
In addition, you cannot select dates and time periods for which Trend history is not
available. For example, if Trend History is available from December 15th, 2011, all dates
prior to this date will be disabled for selection.
If there is no data for a requested date, data is retrieved for the closest date to the
selected date and is indicated in the Trends banner by a blue asterisk.

Quick Reference Guide
Building Advanced Text Queries in Speech Analytics

Speech Analytics provides enhanced search capabilities enabling guided searches for your business issues. This guide is aimed
at advanced users of Speech Analytics and familiarizes users with the sophisticated search capabilities available.
Text queries are used for term searches and for category terms definitions. Building complex text queries requires the use of
operators, which allow you to target specific concepts and to specify retrieval conditions. Using the auto completion and
context-based suggestions features improve query efficiency and can be combined with the use of operators.
Searching with operators refines your query and allows enhanced search capabilities such as:
 Combining words and phrases
 Searching for those terms in the call opening or closing
 Excluding contacts with a certain word
 Applying more importance to contacts with a certain word
 If Speaker Separation has been enabled, searching for a certain term spoken by the agent or by the customer.
IMPORTANT  The following characters cannot be used: Underscore (_), Equal sign (=), Backslash (\), Arrow heads (<>),
Question mark (?)
 Square brackets ([]) can only be used as part of the [START] and [END] operators.
 Operators AND, NOTIN, NEAR, [START], and [END] must be entered in upper case only.
 NOT, if typed in capital letters, is regarded as an operator. Searches with the NOT operator retrieve results where
one term is more dominant than the other. For example, "bill NOT credit" will retrieve contact segments where "bill"
is more dominant than "credit". Contact segment containing "credit" will be assigned lower ranks than results
containing "bill" only. Contact segments where "credit" is more dominant than "bill" will not be retrieved.
 keywords or phrases for which you are searching are not case-sensitive.
 Avoid searching for frequently used words (for example, ¡§hello¡¨ or "yes"). Widely used terms by nature do not
point to specific issues; therefore contacts where these words are found receive lower ranking, which might
prevent them from appearing in search results.
If you want to... Definition Syntax Example Search results include...
Retrieve contacts Enter the word o f interest. account Contacts where the word “account” was used.
containing a word of
interest.
Retrieve contacts Enter your phase enclosed in “close my All contact where the phrase “closes my account” was used. When searching
containing a certain phrase double quotes. account” for a compound word (a word made up of two words, like “checkbook”), use
or compound word the autocompletion list to determine whether “checkbook” has been
(consisting of two words). identified as one word or two.
Retrieve contacts Enter words and phrases of angry annoyed Any contact that contains either the word "angry" or the word "annoyed",
containing any of the your interest. There is no or both.
words or phrases of need to separate the words
interest. with a comma.
Retrieve contacts Enter the words connected angry AND Any contact that contains both the word "angry" and the word "annoyed".
containing all of the words with the AND operator. annoyed
of interest.
Retrieve contacts Use the NOTIN operator credit NOTIN Contacts containing the word "credit" but not when used as part of the
containing a keyword but before the context which is "credit card" phrase "credit card". Note that the search may still return contacts
excluding its usage in irrelevant for your search. containing:
irrelevant contexts. The word "card" when not used as part of the phrase "credit card".
The phrase ¡§credit card¡¨, in case the contacts also contain the word
"credit" not as part of that phrase.
Exclude contacts Add the minus sign (-) credit -card Contacts containing the word "credit", excluding the contacts where the
containing a specific before the word. word "card" was identified. If you are using the minus operator, your query
keyword. must also include at least one other keyword or phrase to search for
(without the minus sign). Queries that include only terms with the minus
sign are not valid and cannot be processed (for example, - card). Do not
use the minus sign (-) in category terms. In a category definition, this
operator is equivalent to adding a term to Avoid Terms.
Retrieve contacts where Enter the words connected close NEAR Contacts containing phrases such as "close my account", "close your
two words appear close to with the NEAR operator. account account" etc. Note that the order of the keywords is significant. In the
each other (from 0 to 3 example entering "close NEAR account" will not return contacts with
words apart). "account" appearing before "close" such as "take into account that if you
close it".
Attribute more importance Add the plus (+) sign bill +credit Contacts where the word "bill" or "credit" is found, but contacts with the
to a certain word. before the word. word "credit" may be ranked higher in the search result list. Do not use the
plus sign (+) in category terms. In a category definition, this operator is
equivalent to adding a term to Very Important Terms.
Quick Reference Guide
If you want to... Definition Syntax Example Search results include...
Retrieve contacts in which Type A: and then the word, A:welcome Contacts where the word "welcome" was spoken by the agent (during the
certain words of phrases phrase or expression you are agent talk time) and also during the talk-over time when the agent and the
were spoken by the agent. looking for, with no space customer were speaking at the same time.
between the colon (:) and
next character.
Retrieve contacts in which Type C: and then the word, C:"thank you" Contacts where the phrase "thank you" was spoken by the customer (during
certain words of phrases phrase or expression you are the customer talk time) and also during the talk-over time when the agent
were spoken by the looking for, with no space and the customer were speaking at the same time.
customer. between the colon (:) and
next character.
To search a call's opening Type [START] and then the [START]welco Contacts where the word "welcome" was used in the call's opening, among
for a specific word or word or phrase of your me the first 100 words of the conversation (around 30 seconds on average).
phrase. interest without a space
between them.
To search a call's closing Type [END] and then the [END]"thank Contacts where the phrase "thank you" was used in the call's closing,
for a specific word or word or phrase of your you" among the last 100 words of the conversation (around 30 seconds on
phrase. interest without a space average).
between them.
Contacts where a specific Inside the [START] operator, [START:60]wel Contacts where the word "welcome" was used in the call opening, among
word or phrase is specify the required number come the first 60 words of the conversation (around 20 seconds on average).
encountered in the call's of words after the colon (:).
opening, among a certain
number of first words.
Contacts where a specific Inside the [END] operator, [END:60]"than Contacts where the phrase "thank you" was used in the call's closing,
word or phrase is specify the required number k you" among the last 60 words of the conversation (around 20 seconds on
encountered in the call's of words after the colon (:). average).
closing, among a certain
number of last words.
Override the default Use the IDF keyword: [IDF] [IDF] “late Contacts containing the phrase “late charge,” and for which the system
search mechanism used before the search entry charge” based its search logic on the IDF weighting, which evalutes how important a
for a single-term search, term is in the index based on the term’s frequency in the index. IDF
forcing the IDF weighting diminishes the weight of terms that occur very frequently, and increases the
to be applied in the search weight of terms that occur rarely.
logic.
Building More Complex Text Queries

By combining operators, words and phrases, you can build more complex and specific text queries, narrowing search results to
the most relevant contact segments only. Use the parenthesis () to group parts of your query. The following are some
examples of more complex queries.
 (call NEAR back) NOTIN "call back today" - this query will retrieve all contact segments where phases similar to "call
you back", "call me back", "didn't not call me back" were detected, excluding the segments where the phrase "call back
today" was found.
 "called before" "called yesterday" "called earlier" - this query will retrieve contact segments where one or more of
these phases were used: "called before", "called yesterday", or "called earlier".
 [END:60]C:"thank you" - with such a text query, the Speech Analytics application will retrieve contact segments where
customers used the phrase "thank you" in the call's closing, among the last 60 words (around 20 seconds on average) of
their part of the conversation.
 [START](how NEAR can NEAR help) - this query will retrieve contact segments where the phrase "How can I help
you?" or similar phrases (for example, "How can we help you?") was used in the call's opening, among the first 100 words
of the conversation (around 30 seconds on average).
 [START]close NEAR account / [START](close NEAR account) - these two queries will retrieve the same results - all
contact segments containing phrases such as ¡§close my account¡¨, ¡§close your account¡¨, ¡§close the account¡¨,
¡§close my recent account¡¨ and similar in the call's opening, among the first 100 words of the conversation.
NOTE  You cannot use parenthesis after Speaker Separation operators (A: or C:). If you want to combine other operators with
A: and C:, place these operators before every operand. For example, a query which reads A:(close NEAR account) is
not valid. You can replace this query with A:close NEAR A:account.
 If you want to use the plus (+) or minus (-) sign before the [START] or [END] operator, you must enclose the operator in
parenthesis. For example, +[START:50]account is not a valid expression. Instead, write +([START:50]account).

Worldwide Headquarters
330 South Service Road
Melville, NY 11747 USA
info@verint.com
1-800-4VERINT
www.verint.com
© 2011 Verint Systems Inc. All Rights Reserved Worldwide.

The contents of this material are confidential and proprietary to Verint Systems Inc.
and may not be reproduced, published, or disclosed to others without express
authorization of Verint Systems Inc.

I360 V11.1 Advanced Speech Analytics User Guide

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

I360 V11.1 Advanced Speech Analytics User Guide

Enviado por

Direitos autorais:

Formatos disponíveis

Advanced Speech Analytics

Confidential and Proprietary Information of Verint Systems Inc.

About this Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Trends and Trend Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Working in the Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3 Customer Interaction Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

4 Contact Search and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Impact 360 Advanced Speech Analytics User Guide 4

Confidential and Proprietary Information of Verint Systems Inc.

Using Autocompletion in Text Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

5 Contact Playback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

6 TellMeWhy Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

7 Category Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

Impact 360 Advanced Speech Analytics User Guide 5

Confidential and Proprietary Information of Verint Systems Inc.

Category Builder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

8 Reports and Data Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

Impact 360 Advanced Speech Analytics User Guide 6

Confidential and Proprietary Information of Verint Systems Inc.

Comparing Two Result Sets Report Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 217

9 Ranking and Document Relevancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

11 Quick Reference Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251

Impact 360 Advanced Speech Analytics User Guide 7

Confidential and Proprietary Information of Verint Systems Inc.

About this Guide

 Product name and version number

Impact 360 Advanced Speech Analytics User Guide 9

Confidential and Proprietary Information of Verint Systems Inc.

Impact 360 Advanced Speech Analytics User Guide 11

Confidential and Proprietary Information of Verint Systems Inc.

Accessing the Speech Analytics Application

Understanding the Speech Analytics Window

Impact 360 Advanced Speech Analytics User Guide 12

Confidential and Proprietary Information of Verint Systems Inc.

Impact 360 Advanced Speech Analytics User Guide 13

Confidential and Proprietary Information of Verint Systems Inc.

Data Analyzed by Speech Analytics

Time Period Covered

Impact 360 Advanced Speech Analytics User Guide 14

Confidential and Proprietary Information of Verint Systems Inc.

Authorized Data and Operations

Speech and Text Analytics

Impact 360 Advanced Speech Analytics User Guide 15

Confidential and Proprietary Information of Verint Systems Inc.

Impact 360 Advanced Speech Analytics User Guide 16

Confidential and Proprietary Information of Verint Systems Inc.

Contact Segment Transcription

Speech Analytics Tasks

Impact 360 Advanced Speech Analytics User Guide 17

Confidential and Proprietary Information of Verint Systems Inc.

Start your Work Session

Impact 360 Advanced Speech Analytics User Guide 18

Confidential and Proprietary Information of Verint Systems Inc.

Contact Playback and Transcription

Impact 360 Advanced Speech Analytics User Guide 19

Confidential and Proprietary Information of Verint Systems Inc.

Reports and Data Export

Impact 360 Advanced Speech Analytics User Guide 20

Confidential and Proprietary Information of Verint Systems Inc.

Speech Analytics Alerts

Speech Analytics Alert Triggers

Impact 360 Advanced Speech Analytics User Guide 21

Confidential and Proprietary Information of Verint Systems Inc.

Speech Analytics Trends Dashboard