Professional Documents
Culture Documents
Seven Keys To Better Forecasting PDF
Seven Keys To Better Forecasting PDF
Forecasting
Mark A. Moon, John T. Mentzer, Carlo D. Smith, and Michael S. Garver
Many companies consider the most important forecasts. This is because system implementation
decisions about forecasting to revolve around the has not been accompanied by effective manage-
selection or development of computer software ment to monitor and control the forecasting pro-
( for preparing the forecasts. They have adopted cess.
the overly simplistic belief that "If we've got good One company we worked with has an excel-
software, we'll have good forecasting." Our re- lent computer system with impressive capabilities
search team, however, has observed numerous of performing sophisticated statistical modeling of
instances of sophisticated computer systems put seasonality and other trends. However, the sales-
into place, costing enormous amounts of time people, who are the originators of the forecast,
and money, that have failed to deliver accurate use none of these tools because they do not un-
Figure 1
The Seven Keys to Better Forecasting
Understand what • Computer system as focus, • Establish forecasting group • An environment in which
forecasting is rather than management • Implement management forecasting is acknowledged
and is not. processes and controls control systems before as a critical business function
• Blurring of the distinction selecting forecasting software • Accuracy emphasized and
between forecasts, plans, and • Derive plans from forecasts game-playing minimized
goals • Distinguish between forecasts
and goals
Forecast demand, • Shipment history as the basis • Identify sources of information • Improved capital planning
plan supply. for forecasting demand • Build systems to capture key and customer service
• "Too accurate" forecasts demand data
Eliminate islands • Mistrust and inadequate infor- • Build a single "forecasting • More accurate, relevant, and
of analysis. mation leading different users infrastructure" credible forecasts
to create their own forecasts • Provide training for both users • Optimized investments in
and developers of forecasts information/communication
systems.
Use tools wisely. • Relying solely on qualitative • Integrate quantitative and • Process improvement in
or quantitative methods qualitative methods efficiency and effectiveness
• Cost/benefit of additional • Identify sources of improved
information accuracy and increased error
• Provide instruction
Make it important. • No accountability for poor • Training developers to under- • Developers taking forecasts
forecasts stand implications of poor seriously
• Developers not understanding forecasts • A striving for accuracy
how forecasts are used • Include forecast performance • More accuracy and credibility
in individual performance
plans and reward systems
Measure, measure, • Not knowing if the firm is • Establish multidimensional • Forecast performance can be
measure. getting better metrics included in individual perfor-
m Accuracy not measured at • Incorporate multilevel mance plans
relevant levels of aggregation measures • Sources of errors can be
• Inability to isolate sources of • Measure accuracy whenever isolated and targeted for
forecast error and wherever forecasts are improvement
adjusted • Greater confidence in forecast
process
the company. We believe the sales forecast and was not fulfilled. Mechanisms are needed to allow
the sales goal must be distinct, because the be- salespeople to provide valuable information about
haviors they are meant to influence can conflict. customers who would order more if they could.
In addition, records of orders accepted but not
Key #2: Forecast Demand, Plan Supply filled in the period demanded adds to the de-
mand versus supply level of information. Finally,
One mistake many companies make is forecast- such electronic data interchange (EDI) informa-
ing their ability to supply goods or services rather tion as point-of-sale (POS) demand, retail inven-
than actual customer demand. At the beginning tory levels, and retailer forecasts are all valuable
of the forecast cycle, it is important to create sources of information that help a company
predictions that are not constrained by the firm's move toward demand forecasting.
capacity to produce. Consider the forecaster for a Although it is more difficult, forecasting true
certain product who questions the company's demand will help a company make sensible,
sales force and learns they could sell 1,500 units long-term decisions that can profoundly affect its
per month. At the same time, current manufactur- market position. By identifying where capacity
ing capacity for that product is 1,000 units per does not meet demand forecasts, the company
month. If the forecaster takes that production has valuable information on where to expand
capacity into account when creating initial fore- capacity through capital planning. Such a long-
casts, and predicts 1,000 units, there is no record term program of matching capacity planning to
of the unmet demand of 500 units per month, forecasts will reduce the incidence of chronic
and the information on where to expand manu- underforecasting and result in higher levels of
facturing capacity is lost. customer satisfaction.
This problem often occurs when historical
shipments are used as the basis for generating Key #3: Communicate, Cooperate,
forecasts. Forecasting shipments will only predict Collaborate
a company's previous ability to meet demand.
Suppose demand for a particular product in the Companies that forecast most effectively consider
past had been 10,000 units per month, but the it critical to obtain input from people in different
supplier could only ship 7,500. Corporate history functional areas, each of whom contributes rel-
would show shipments at 7,500 units per month, evant information and insights that can improve
( thus causing this amount to be projected and overall accuracy. But employees are often unable
produced again the following month. The result or unwilling to work across functions to achieve
is twofold: the impression of an accurate forecast- high levels of forecasting performance. To do so
ing system, but an actually recurring unfulfilled requires a great deal of communication across
monthly demand of 2,500 units. Forecasting department boundaries, and not all communicat-
based on shipping history only leads a company ing is equal; some companies are simply better at
to repeat its former mistakes of not satisfying it than others.
customer demand. Predicting actual demand When it comes to cross-functional forecast-
allows measurement of the disparity between i ng, we distinguish among three levels: communi-
demand and supply so it can be reduced in fu- cation, cooperation, and collaboration. Compa-
ture periods through plans for capacity expan- nies at lower levels of sophistication merely com-
sion. municate. This can take the form of one-way
Often the symptom of this key is the attitude, reports, in which one department responsible for
" We do a great job of forecasting. We are very forecasting informs other functional areas of the
accurate, always selling close to what we fore- results of its efforts. With coordination, represen-
cast." Notice in the previous example that the tatives from different functional groups meet to
forecast accuracy would appear very good be- discuss the forecast. Often, however, one area-
cause both the forecast and the actual sales were usually the one that "owns" the forecast-will
7,500 units each month. The key, however, is the dominate the discussions and work to persuade
failure to realize the 2,500 units in sales lost each the other functions to accept the forecast it has
month because of an inability to meet demand. created.
In fact, the "true" demand forecasting accuracy Coordination is superior to one-way commu-
was not 100 percent, but only 75 percent. Fore- nication, because at least there is opportunity for
casting by shipments and obtaining accurate re- some dialogue. But it does not promote as effec-
sults are often symptomatic of chronic underfore- tive a forecasting process as when different con-
casting of demand. stituencies in a company collaborate. Here, the
Unfortunately, determining actual customer views of each functional area receive equal con-
demand is more difficult than predicting a com- sideration, and no one department dominates.
pany's ability to supply. Systems and processes Such collaboration is most likely to occur when
are needed to capture this elusive demand that management of the forecasting process resides in
A
sures are not used to evaluate a person's job s we work with companies, many of
performance, or do not provide support for iden- them come to realize what a profound
tifying sources of forecasting error. impact these seven keys can have on
Effective measures evaluate accuracy at dif- their sales forecasting practices. As they improve
ferent levels of aggregation. Logistics operations those practices, they experience reductions in
are interested in forecast performance at the costs and increases in customer and employee
SKUL level; sales managers may be more inter- satisfaction. Costs decline in inventory levels, raw
ested in a forecast stated in dollars and at the materials, production, logistics, and transporta-
territory or product line level of aggregation. tion. Greater customer satisfaction accrues from
Performance metrics should support these vari- more accurately anticipating demand and. subse-
ous units of measure as well as the aggregation quently, fulfilling that demand more often. Greater
of demand at different levels. employee satisfaction comes from a more under-
It is also important to track accuracy at each standable process, easier information access and
point at which forecasts may be adjusted. As an transfer, and explicit rewards tied to performance.
illustration, the forecasting task of the sales force But the first step any company must take before
of one company is to examine "machine-gener- realizing these benefits is to recognize the impor-
ated" forecasts for their customers and make tance of sales forecasting as a management func-
adjustments. Those adjustments are then mea- tion. With this recognition comes a willingness to
sured against actual sales to determine whether commit the necessary resources to improving this
the salesperson's adjustment improved the fore- critical process. O
cast or not. Similarly, the product manager's job is
to take the machine-generated forecast, which References
has been adjusted by the sales force, and make
further adjustments based on a knowledge of Kenneth B. Kahn and John T. Mentzer, "The Impact of
market conditions or upcoming promotional Team-Based Forecasting," Journal of Business Forecast-
events. Once again, these adjustments are mea- ing, Summer 1994, pp. 18-21.
sured against actual sales to determine whether
they improved the forecast. In both cases, the Kenneth B. Kahn and John T. Mentzer, "Forecasting in
salespeople and the product manager gain feed- Consumer and Business Markets," Journal of Business
Forecasting, Summer 1995, pp. 21-28.
back that helps them improve their efforts.
Finally, companies should assess forecasting John T. Mentzer and Carol C. Bienstock, Sales Forecast-
accuracy in terms of its impact on business per- ing Management (Thousand Oaks, CA: Sage Publica-
formance. Accurate forecasts should not be an tions, 1998).
end in themselves, but rather a means to achiev-
ing the end, which is business success. Improve- John T. Mentzer and James E. Cox, Jr., "A Model of the
ments in accuracy require expenditures of re- Determinants of Achieved Forecast Accuracy. - Journal
sources, both human and financial, and should of Business Logistics, 5, 2 (1984a): 143-155.
be approached in a return-on-investment frame-
work. For example, in a distribution environ- John T. Mentzer and James E. Cox, Jr., "Familiarity,
Application, and Performance of Sales Forecasting
ment, maintaining or improving customer service Techniques," Journal of Forecasting, 3 (1984b): 27-36.
may be a worthy corporate objective. Investment
in more accurate forecasts may be one way to John T. Mentzer and Kenneth B. Kahn, "Forecasting
achieve that objective. However, if the investment Technique Familiarity, Satisfaction, Usage, and Applica-
required to improve accuracy significantly is very tion, "Journal of Forecasting, 14, 5 (1995): 465-476.