Escolar Documentos
Profissional Documentos
Cultura Documentos
Retail
Reward Customer Loyalty
with SAP Predictive Analysis
and SAP Business Objects
Dashboards
DOCUMENT CLASSIFICATION: INTERNAL AND PARTNER
General Information:
Authors:
I817248
I304624
April 2014
Page 1
Page 2
TABLE OF CONTENTS
1.
2.
1.2.
1.3.
Protagonists ............................................................................................................................................ 4
1.4
Prerequisites/Restrictions........................................................................................................................ 5
2.2.
2.3.
3.
2.2.1.
2.2.2.
2.2.3.
Users .......................................................................................................................................... 5
2.2.4.
3.2
Please select for which audience this demo script is intended by selecting the appropriate boxes below.
Retailers
CPG
1.3. PROTAGONISTS
Marketing Analyst/Managers
TECHNICAL REQUIREMENTS
2.1 PREREQUISITES/RESTRICTIONS
The software licensing requirements are needed to implement this solution are:
SAP BusinessObjects Enterprise 4.0
SAP BusinessObjects Dashboard Design (Xcelsius)
SAP BusinessObjects Web Intelligence
SAP Predictive Analysis
SAP JAM
Please download and install the latest SAP BusinessObjects Mobile App:
Launch App Store and search SAP BusinessObjects Mobile and you will see the App
Install the app
2.2
2.2.1
System Landscape
SAP Demo Cloud
2.2.2
System Access
This demo is available from the SAP Demo Cloud, in the following showrooms:
-
For specific details on how to connect to the showrooms above (including mobile setup), please refer to the
documentation: SAP Demo store, demo 9169:
https://www.sapdemostore.com/scenario/9169
For SAP Demo cloud generic information (access, registering, set up SDC connector, etc.) please go to the SAP Demo
Store documentation How to Access the SAP Demo Cloud?
https://www.sapdemostore.com/scenario/8024
2.2.3
Languages Supported
Please select the standard languages in which this demo is available (i.e. translated into).
English
German
French
Spanish
Portuguese
Japanese
Simplified Chinese
Korean
Software Component
SAP Predictive Analysis
SAP BusinessObjects Mobile
Release
Support Pack
1.0.9
4.4
DEMO SCRIPT
3.1 STORY
1) It is close to quarter end, Jack Lee, Operations Director at BestRun Retailer, notices from his Dashboard that
sales will not meet target would like to run some promotions to increase sales. They would like to identify the
most profitable customers and their purchase preference, in order to reward their loyalty and promote crosssell opportunities.
2) Lead Analyst, Mike starts with high level visualization of where their customers are and their demographic
information.
3) To gain more insight, Mike uses Clustering to perform customer segmentation. He discovers a small cluster of
customers that had made relatively high amount and item in purchases. They are also younger, more affluent
customers.
4) Mike zooms in this specific group of customers, he also notices that most of them made last purchase only 1
week ago. This looks like the loyal, profitable customers they want to target. Mike also discovers that these
customers make purchases online.
5) Mike exports the segmentation result to a CSV file.
6) Mike now needs to figure out what items can be offered in the promotion to this group of customers. He
conducts an association run for market analysis. Mike concludes that focusing on the cross-sell relationship
between these items would lead to the a win-win situation for both Best Run and its loyal customers.
7) Best Run sends special promotional online offers to this group of customers. The promotion is a success,
along with other actions that management has taken, Best Run closes out the quarter above target. In
addition, due to analysis done in SAP Predictive Analysis, Best Run is now able to track their top customers s
KPI on a ongoing basis. Best Run is also able to create individual view of their most loyal customers activities
and metrics. This dashboard can also viewed on a mobile device. Team members can collaborate using SAP
Jam to quickly create personalized offers.
3.2
3.2.1
Step 1.0 Download and install SAP BusinessObjects Mobile App on iPad
What To Do
3.2.2 Demo
What to say
It is close to quarter end, It is close to quarter end, Jack Lee, Operations Director at Best Run, reviews his Sales Performance Dashboard. He
notices that Total Sales is not going to meet target.
Though HANA is not directly used for tstoring the back end data displayed on these dashboard, you can easily talk here about what HANA can bring
to your customers/prospects
When used as a transactional or Datawarehousing platform, SAP HANA provides on-the-fly analysis for all combinations of data. You can make
immediate decisions using real-time operational analytics whether the data comes from your SAP applications, third-party solutions, or custom
applications.
Big Data Warehousing:
SAP HANA unleashes the potential of Big Data with its ability to handle large volumes and a variety of structured and unstructured data in real-time.
Build a data warehouse integrated with Hadoop or easily migrate your existing SAP NetWeaver Business Warehouse (BW) to SAP HANA and make
decisions within the window of opportunity.
What To Do
Save swf files customer
loyalty.swf and customer
Analtycis.swf to your local drive.
Jack notices the red alert signals next to the Total sales and net sales metrics. It seems that sales for this quarter
is not meeting target.
Lava Dashboard works on mobile devices too. When Jack is outside of his office, he can easily review the same
dashboard on his iPad and make the same observations.
smithjo/welcome
Tap on
What to say
Jack realizes that they need to run some promotions to increase sales.
One of the action items is to identify the most profitable customers and their purchase preference, in order to reward their loyalty and promote cross-sell
opportunities. Best Run will utilize SAP Predictive Analysis.
Use also this opportunity to discuss SAP HANA advantages for PA:
SAP HANA is a fully ACID compliant, in-memory, columnar, massively parallel processing platform that provides common database for online transactional
processing (OLTP) and online analytical processing (OLAP), eliminating redundancy and latency. It also unifies end-to-end data processing with search,
text, geospatial and predictive analytics on top of the in-memory columnar foundation, further extending the benefit of redundancy and latency elimination to
all data processing workloads. SAP HANA provides a powerful suite of predictive, spatial, and text analytics librairies that can run across multiple data
sources. Visualize new opportunities and gain deep insights with unified analysis of all of data types.
SAP HANA also enables real-time prescriptive analytics by reducing layers of processing and accelerating complex computations required for scheduling
and simulations.
What To Do
Start SAP Demo Cloud CSA
(Green Phoenix) Showroom and
Click on Desktop.
Click on
Mike uses data from customer table, which contains customer demographic information gathered from loyalty card
program and transcations information. He can view the Facets view of our data, which provides the frequencies of the
different values for our variables.
Click on
icon next to
Income Column.
In Facet view, Mike can directly sort to view the values of each varialbe. For instance, he click on the sort button for
Income and notices that the highest income of the customers is $307064 and there is one customer with that amount of
income.
Click on
Region
Predictive Analysis has visualization and graphing capabilities built right into the modelling tool, so there is no need to
move between programs when Mike conducts his analysis. Mike would like to perform some high level analysis first.
He starts with a Geo chart to gain a high level view of the customers.
next to Dimension
Click on Confirm.
Click on OK
SAP AG 2014 / SCENARIO ID: 8860
Page 18
Mike is able to tell right away that California and Midwest states have the highest number of customers. And he also
notices the states with very light colors, these are the states with fewer customers.
Mike can get an additional layer view of the data quickly by adding Time last purchase made to the same chart.
Mike can share this visualization in one click via email with his colleagues.
The Geo map chart provides a high level overview; but does not provide the in depth analysis needed to identify the most
loyal customers. To conduct advanced analysis, Mike moves to the Designer area of SAP Predictive Analysis to set up
the workflow needed for his analysis.
First, Mike would like to run a Clustering Algorithem to segment the customers.
And click on
Click on
on top right of window to run analysis.
He starts the analysis and clicks Yes to proceed to view the result.
And then proceed to Visualization area when he receives the run success window.
Click on
icon the
bottom corner of the Result view.
Customers are segmented into 8 groups. Mike notices clusters 1, 2, 6 and 8, these clusters all have high density, which
means customers within each of these groups share very similar characteristics.
Mike observes that cluster 1 consists only of the younger age group.
Click on
Mike now switches to Visualize mode to get a better understanding of the segmentation result.
He would like to know the purchasing pattern for each cluster. He starts with a bubble chart.
Mike notices the big bubble that is apart from the rest of bubbles. Based on the legend, this bubble represents cluster 1
customers. It seems that cluster 1 customers have the highest lifetime purchase amount, and highest number of total
items purchased in the last 6 months.
Note: Cluster number is assigned randomly when you run this algorithm. It might not be cluster #2 when you
run this in PA. Please note the cluster with the least number of records; this is the cluster that will be used in
the following steps.
Next, Mike introduces some demographic variables to the analysis and discovers that cluster 1 customers are a younger
group of customers with high credit scores. This is definitely the type of customers that retailers would like to target.
Next, Mike conducts analysis on this cluster customer alone. First, he creates a tag cloud chart to check their buying
frequency and most recent buying period. It turns out most of them made a purchase pretty recently, ranging from 1
week ago to 6 month ago. And they purchased relatively high amount of popular items as indicated by the color
intensity of the words. It seems that Mike has found the loyal customers he is searching for. However, Mike also
noticed that more customers in this cluster made last purchase 1 month ago, so they could be at the verge of
abandonment.
Now that Mike has found the loyal customers he is searching for, he needs to understand how they purchase. Mike uses
a pie chart and discovers that 77% of them shopped online. This fits the younger, affluent customers shopping pattern.
Click on
Tab
Mike would like to export the segmentation result to a file. He chooses to export as a CSV file. This output assigns
cluster number to each customer record.
Click on
tab
Click on
tab.
Now Mike needs to decide what items can be used to reward this group of loyal customers and address the potential
abandonment issue. He conducts a R-Apriori Association analysis for basket analysis. The apriori algorithm will output a
set of rules representing cross-sell or up-sell relationships. First, Mike selects the item column to be product, as he
wants to start by conducting his analysis at the product level. Next, Mike selects transaction ID column to be trans.
Support is the minimum percentage of transactions in our data that need to contain items from the categories that we
are analyzing. If we are looking at the purchase relationship between categories A and B with a support level of 5%, then
greater than 5% of the purchases must contain products from categories A & B. Mike sets this value at 0.05.
Confidence is the liklihood that a customer will purchase from category B given they have purchased A, in essence what
percent of transactions that contain A also contain B. Mike sets this value at 0.1. Now he can save and run his market
basket analysis.
Mike will get a notification that his analysis has been successfully executed, and he will be able to see the relationships
between his purchases.
Now he can evaluate the results of our analysis with the provided purchase relationships. The PreRule product
categories on the far left are the traffic driver products. The PostRule product categories are affinity products, basket
builders that are most often purchased with something else. The Lift value on the far right is the one number used to
evaluate these relationships as cross-sell or up-sell opportunities, the higher the better. Generally a Lift value over 1 is
considered a viable opportunity.
The basket analysis results suggests that customers usually buy collection pencil skirt when they shop for chiffon tops
and suede pumps. Mike concludes that focusing on the cross-sell relationship between these items would lead to the a
win-win situation for both Best Run and its loyal customers.
Since Mike has exported the segmentation result already, in the output table, each record is assigned a cluster number.
He can easily identify cluster #1 customer and email them an online special offer of buy 1 chiffon top, get 1 collection
pencil skirt 75% off.
The promotion is a success, along with other actions that management has taken, Best Run closes out the quarter above
target. In addition, due to analysis done in SAP Predictive Analysis, Best Run is now able to track their top customers s
KPI on a ongoing basis.
Click on TOP
In addition, due to analysis done in SAP Predictive Analysis, Best Run is now able to track their top customers s KPI on
a ongoing basis. They have created a Top filter specifically to track metrics related to this group of customers.
Best Run is also able to create individual view of their most loyal customers activities and metrics. And take immediate
action to retain/reward these customers. Team members can collaborate using SAP Jam to quickly create personalized
offers.