Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract
Continuous improvement in sales forecasting is a worthy goal for any organization. This paper describes a methodology for conducting a
sales forecasting audit, the goal of which is to help a company understand the status of its sales forecasting processes and identify ways to
improve those processes. The methodology described here has been developed over a 5-year period, involving multiple auditors, and has
been implemented (to date) at 16 organizations. This methodology revolves around three distinct phases: the as-is phase, in which the audit
team seeks to understand fully a companys current forecasting process; the should-be phase, in which the audit team presents a vision of
what world-class forecasting should look like at the audited company, and the way-forward phase, in which the audit team presents a
roadmap of how the company can change its forecasting processes to achieve world-class levels. Those companies that have responded
positively to the audit process have experienced significant improvement in their forecasting performance. The paper concludes by
presenting lessons from audits conducted to date, as well as implications for management practice and future research.
2002 International Institute of Forecasters. Published by Elsevier Science B.V. All rights reserved.
Keywords: Forecasting practice; Audit; Forecasting management; Performance measurement; Forecasting systems; Supply chain
1. Introduction
Forecasting has been consistently recognized as an
important capability for business planning and management (Armstrong, 1987; Cox, 1987, 1989; Fildes
& Hastings, 1994; Makridakis & Wheelwright, 1977;
Mentzer & Gomes, 1994; Sanders & Manrodt, 1994;
Wright, 1988). Regardless of industry, or whether the
company is a manufacturer, wholesaler, retailer, or
service provider, effective demand forecasting helps
organizations identify market opportunities, enhance
channel relationships, increase customer satisfaction,
*Corresponding author. Tel.: 11-865-974-8062; fax: 11-865974-1932.
E-mail address: mmoon@utk.edu (M.A. Moon).
reduce inventory investment, eliminate product obsolescence, improve distribution operations, schedule
more efficient production, and anticipate future
financial and capital requirements (Galfond,
Ronayne, & Winkler, 1996; McIntyre, Archabal, &
Miller, 1993).
To improve forecasting, companies increasingly
take advantage of computer and information system
technologies. Point of sale (POS) data collection is
gathering near-real-time sales movement at a stock
keeping unit by location (SKUL) level of detail
(Smart, 1995). Electronic data interchange (EDI) is
used to transmit detailed sales information throughout various marketing channels to product distributors and manufacturers (Mentzer & Kahn,
1997). Forecasting systems that were once housed on
0169-2070 / 02 / $ see front matter 2002 International Institute of Forecasters. Published by Elsevier Science B.V. All rights reserved.
PII: S0169-2070( 02 )00032-8
Mentzer and Kent (1999), and Mentzer and Schroeter (1994). These articles suggest that forecasting
implementation entails more than the application of
more accurate forecasting techniques and that effective forecasting management, as with effective system implementation, may lead to improved operating
and business performance.
Three exemplary articles from this stream of
research bear elaboration. The first (Armstrong,
1987) presented an idealized forecasting case study
and, from this case, derived 16 pitfalls in forecasting. From these pitfalls, Armstrong developed a
forecasting audit checklist for companies to review
their own management processes. This audit included the steps: (1) assess the methods without the
forecasts, (2) assess the assumptions and data used in
the forecast, (3) assess the uncertainty of the forecast, and (4) assess the costs of the forecast. Although not based upon an actual company, this paper
did establish the importance of auditing company
forecasting processes instead of just methods.
The second paper (Fildes & Hastings, 1994) drew
upon the organizational behavior and diffusion literature to develop a theoretical model of a marketing
forecasting system. From this model, a series of
theoretical statements were derived in the areas of
(1) the forecaster and the decision maker, (2)
information flows, and (3) technical characteristics
of the forecast. The authors then tested their theoretical statements in a case study of a company with 10
separate divisions and concluded that:
techniques and forecasting software should meet
the needs of the forecaster;
with limited time and technical expertise, implementation will also critically depend on the
database system and the supporting organization
design;
without accountability for forecast improvements
and adequate time and resources, little will happen; and
forecast improvement depends on organizational
design.
The importance of Fildes and Hastings (1994)
research to this paper is the establishment of a
theoretical basis from which to develop a model of
forecasting management, and the use of case companies to test the model.
In the third article, Schultz (1984) discussed
managerial issues surrounding the successful implementation of new forecasting models. He outlined
12 implementation profile factors which he summarized from his own and other forecasting management research. These factors include top management support, the relationship between forecast users
and model designers, goal congruence, implementation strategy and resources, and costbenefit justification. The importance of the Schultz (1984) paper
is its explicit recognition that even the best forecasting models will not affect overall corporate performance without attention to managing organizational change processes.
While these studies have provided valuable information and insight into forecasting management,
they have not offered a generalizable framework that
may be used as a prescriptive tool to improve
forecasting management. As Armstrong (1988)
pointed out, a key area where researchers have yet to
provide assistance to practitioners is in implementation. In other words, how can practising forecasters
take the knowledge that has been developed about
forecasting, and put it into action to improve forecasting in their particular organization? It is this
question that the remainder of this paper addresses.
whether or not their organizations forecasting processes are good or bad, it is useful to have a standard
against which those processes can be compared.
Armstrong (1987) provided such a standard with his
16-point forecasting audit checklist. This checklist
was divided into the dimensions of forecasting
methods, assumptions and data, uncertainty, and
costs and benefits (see Table 1).
Fildes and Hastings (1994) utilized organizational
behavior and diffusion theory to present a normative
model of the Marketing Forecasting System. Based
upon this model, the authors suggested evaluating
the forecaster and the decision maker, information
flows, and the technical characteristics of the forecast
(see Table 1).
Mentzer et al. (1999) based their four-dimensional
framework on the work of Armstrong (1987), Fildes
and Hastings (1994), and Schultz (1984), and the
findings from a 15-year, three-phase research program in forecasting management. Phases one (Mentzer and Cox, 1984a) and two (Mentzer and Kahn,
1995) were based upon two large survey-based
quantitative analyses of sales forecasting practices.
Phase three involved a benchmark study of the
forecasting practices at 20 leading companies representing a variety of industries. Based on in-depth
analyses of company processes and documents, as
well as interviews with both users and developers of
forecasts, Mentzer et al. (1999) proposed that, in
order to adequately understand the overall management of the forecasting process in a company, that
process must be investigated along the following
four dimensions.
1. Functional integration, concerned with the role of
collaboration, communication, and coordination
of forecasting management with the other business functional areas of marketing, sales, finance,
production, and logistics.
2. Approach, concerned with which products and
services are forecast, the forecasting techniques
used, and the relationship between forecasting and
planning.
3. Systems, addressing the evaluation and selection
of hardware and software combinations to support
the sales forecasting function as well as the
integration of forecasting systems with other
planning and management systems in the organization.
Table 1
Forecasting management frameworks
Fildes and Hastings (1994)
Forecasting methods
1.
Forecasting independent of top management?
2.
Forecast used objective methods?
3.
Structured techniques used to obtain judgments?
4.
Least expensive experts used?
5.
More than one method used to obtain forecasts?
6.
Users understand the forecasting methods?
7.
Forecasts free of judgmental revisions?
8.
Separate documents prepared for plans and forecasts?
Functional integration
1.
Degree of communication, coordination, and
collaboration between forecasting group and
other functional areas
2.
Organizational location of the forecasting group
3.
Existence and form of consensus forecasting
meetings
4.
Recognition of forecasting needs of various
functional areas
5.
Accountability/performance rewards for personnel
involved in developing the forecasts
Information flows
4.
Information flow from the environment
5.
Intraorganizational information flows and loss
of information
6.
IT support for information flows
Technical characteristics of the forecast
7.
Accuracy and bias
8.
Responsiveness and speed
9.
Uncertainty estimation
Approach
6.
7.
8.
9.
10.
11.
Systems
12.
13.
Armstrong (1987)
10
4. Performance measurement, considering the metrics used to measure sales forecasting effectiveness and its impact on business operations.
In addition to identifying these four dimensions of
forecasting management, Mentzer et al. (1999) articulated four stages of effectiveness within each
dimension (see Figs. 14). Their article provided a
detailed description of the characteristics that can be
found at each of the four stages of effectiveness
within each of the four dimensions. From the Table 1
comparison of the Armstrong (1987), Fildes and
Hastings (1994), and Mentzer et al. (1999)
11
correspond to similar criteria under functional integration; information flows correspond to both
functional integration and systems; and technical
characteristics of the forecast correspond to much in
the approach category.
12
13
14
Table 2
Themes across sales forecasting management frameworks a
Themes
Exemplars
MBK
F&H
Armstrong
Organization
Communications
Coordination
Location
Interaction
Link to planning and
decisions
1
1
2
3
4, 5, 6
1, 8
From environmental
Within supply chain
Intra-organization
Availability
Information systems
12
8
12
13
14
Top-down / bottom-up
Techniques
Product segmentation
Assumptions explicit
Measurement of accuracy
Impact of external factors
Reasons for forecast
uncertainty explored
Variables forecast and
lead time
7
10
9
Information
Technical issues
The forecaster
and users
Costs and
benefits
a
Training
Incorporation of judgement
Forecasters style
Users knowledge
Resources available
Value (perceived and actual)
to organization
15
16
4
4
5
6
10
2, 3, 4, 5, 7, 11
7
9
15
13
14
12
11
10
2
1
6
17
9
16
The authors wish to thank Professor Robert Fildes for the original idea and guidance to develop this table.
15
Table 3
Characteristics of audited companies
Company
Consumer?
Allied Signal
Avery Denison
ConAgra
Corning
DuPont
Eastman Chemical
Ethicon
Exxon
Hershey USA
Lucent Technologies
Michelin North America a
Motorola b
Pharmavite
Smith & Nephew
Union Pacific RR
WilliamsonDickie
X
X
X
Totals
8 (50%)
a
b
Business
to business?
Primarily
direct
sales?
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Primarily
sales through
distributors?
X
X
X
X
X
10 (63%)
6 (38%)
12 (75%)
16
component of validity and reliability in interviewbased research (Armstrong, Gosling, & Weinman,
1997; Griggs, 1987; Hughes & Garrett, 1990;
Kurasaki, 2000). Each interview is audiotaped so any
differences in interpretation can be resolved at the
time of data analysis. Interviews typically include
one subject and two auditors, although occasionally
two or more subjects are present during a single
interview.
The interview typically begins with one member
of the audit team briefly describing the audits
purpose, giving assurances of confidentiality, and
obtaining permission to audiotape the conversation.
The auditors then ask the subject to describe his or
her role in the forecasting process. Probes include
questions such as what do you do with that information?, where do you obtain that information?, and is the information you obtain from that
source credible and complete? While a considerable
portion of the interview is spent gaining an understanding of the subjects formal role and responsibility with regard to the forecasts, the auditors also
probe to obtain insights as to the subjects satisfaction with the sales forecasts and the sales forecasting
process. Focus is placed on understanding any
frustrations or problems the subject has with the
forecasting process and his or her role in that
process. Each interview ends with a wish-list
question, where one of the auditors asks the subject
to describe what he or she would do to make the
forecasting process more effective. Interviews typically last 45 min to an hour.
At the 16 companies audited to date, the number
of interviews has ranged from 22 to 64, with an
average of 32 per company. Following completion of
the interviews, the auditors combine all interview
notes, then distribute those notes to all audit team
members. With data collection completed, data analysis begins.
17
18
19
20
21
22
23
24
References
Armstrong, D., Gosling, A., & Weinman, J. (1997). The place of
inter-rater reliability in qualitative research: an empirical study.
Sociology, 31 (3), 597606.
Armstrong, J. S. (1987). The forecasting audit. In Makridakis, S.,
& Wheelwright, S. C. (Eds.), The handbook of forecasting.
New York: John Wiley, pp. 584602.
Armstrong, J. S. (1988). Research needs in forecasting. International Journal of Forecasting, 4, 449465.
Armstrong, J. S., Brodie, R. J., & McIntyre, S. H. (1987).
Forecasting methods for marketing. International Journal of
Forecasting, 3, 355376.
Arter, D. R. (1989). Evaluate standards and improve performance
with a quality audit. Quality Progress, 22 (9), 4143.
Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (1999). The
bullwhip effect: managerial insights on the impact of forecasting and information on variability in a supply chain. In
Quantitative models for supply chain management. Boston,
MA: Kluwer Academic.
Churchill, Jr. G. A., Ford, N. M., & Walker, Jr. O. C. (1993). In
4th ed, Sales force management. Homewood, IL: Richard D.
Irwin.
Closs, D. J., Oaks, S. L., & Wisdo, J. P. (1989). Design
requirements to develop integrated inventory management and
forecasting systems. In Annual conference proceedings. St.
Louis, MO: Council of Logistics Management.
Cox, J. E. (1987). An assessment of books relevant to forecasting
in marketing. International Journal of Forecasting, 3, 515
527.
Cox, J. E. (1989). Approaches for improving salespersons
forecasting. Industrial Marketing Management, 18, 307311.
Dew, J. R. (1994). The critical role of auditing in continuous
improvement. National Productivity Review, 13, 417422,
Summer.
Fildes, R., & Hastings, R. (1994). The organization and improvement of market forecasting. Journal of the Operations Research Society, 45 (1), 116.
Fildes, R., & Ranyard, J. C. (1997). Success and survival of
operational research groupsa review. Journal of the Operational Research Society, 48 (4), 336360.
25