The Forecasting Process

Many field sales managers are guilty of plotting Problem Definition

their historical sales data and computing sales forecast
for months without considering the assumptions and the
type of model they are using. Unfortunately, the sales
forecasting process requires a little more careful Gathering Information
consideration than that. Before building a model there
are some key questions to ask and steps to follow:

 What type of market am I working with? Choosing and fitting

 Is there precedence to what I’m doing? models
 Is the data I’ve extrapolated accurate?
 Have I tested different models?
 Which model best fits my sales process? Using and evaluating
a forecasting model
Insufficient answers to questions like these may
lead to picking an inappropriate sales forecasting process for your sales management process
model and leave your predictions well off the mark.

To avoid all that I’ve gone ahead and put together a rough guide on what I believe are the
key steps and put them in chronological order.

1. Problem Definition

Often this is the most difficult part of forecasting. Defining the problem carefully requires
an understanding of the way the forecasts will be used, who requires the forecasts, and how the
forecasting function fits within the organization requiring the forecasts. A forecaster needs to
spend time talking to everyone who will be involved in collecting data, maintaining databases,
and using the forecasts for future planning.

2. Gathering Information

There are always at least two kinds of information required: (a) statistical data, and (b)
the accumulated expertise of the people who collect the data and use the forecasts. Often, it will
be difficult to obtain enough historical data to be able to fit a good statistical model.

Occasionally, old data will be less useful due to structural changes in the system being
forecast; then we may choose to use only the most recent data. However, remember that good
statistical models will handle evolutionary changes in the system; don’t throw away good data
unnecessarily.

Page 1 of 11
 3. Choosing and Fitting Models

The best model to use depends on the availability of historical data, the strength of
relationships between the forecast variable and any explanatory variables, and the way in which
the forecasts are to be used. It is common to compare two or three potential models. Each model
is itself an artificial construct that is based on a set of assumptions (explicit and implicit) and
usually involves one or more parameters which must be estimated using the known historical
data.

4. Using and Evaluating a Forecasting Model

Once a model has been selected and its parameters estimated, the model is used to make
forecasts. The performance of the model can only be properly evaluated after the data for the
forecast period have become available. A number of methods have been developed to help in
assessing the accuracy of forecasts. There are also organizational issues in using and acting on
the forecasts.

Data Consideration

Explanatory vs. Time Series Forecasting

 Explanatory models
 Assume that the variable to be forecasted exhibits an explanatory
relationship with one or more independent variables

 Time series forecasting

 Makes no attempt to discover the factors affecting its behavior. Hence
prediction is based on past values of a variable. The objective is to
discover the pattern in the historical data series and extrapolate that
pattern into the future.

Data Patterns

• The data that are used most often in forecasting are time series.
• Time series data are collected over successive increments of time.
 Example: Monthly unemployment rate, The quarterly gross domestic
product, the number of visitors to a national park every year for a 30-year
period.

• Such time series data can display a variety of patterns when plotted over time.

Page 2 of 11
 A time series is likely to contain some or all of the following components:

• Trend
• Seasonal
• Cyclical
• Irregular

Trend:

Trend in a time series is the long-term change in the level of the data i.e., observations
grow or decline over an extended period of time.

 Positive trend
• When the series move upward over an extended period of time
 Negative trend
• When the series move downward over an extended period of time
 Stationary
• When there is neither positive or negative trend.

Seasonal:

Seasonal pattern in time series is a regular variation in the level of data that repeats itself
at the same time every year.

Data is affected by weather, vacation, and holidays and usually consists of periodic,
repetitive, and generally regular and predictable patterns

Seasonality can repeat on a weekly, monthly or quarterly basis, these periods of time are
structured and occur in a length of time less than a year.

Examples:
 Retail sales for many products tend to peak in November and December.
 Housing starts are stronger in spring and summer than fall and winter.

Cyclical:

Cyclical patterns in a time series are presented by wavelike upward and downward
movements of the data around the long-term trend.

Page 3 of 11
 A cyclic pattern exists when data exhibit rises and falls that are not of fixed period. The
duration of these fluctuations is usually of at least 2 years.

• They are of longer duration and are less regular than seasonal fluctuations.
• The causes of cyclical fluctuations are usually less apparent than seasonal
variations.

Model Selection

The pattern that exists in the data is an important consideration in determining which
forecasting techniques are appropriate. To forecast stationary data, use the available history to
estimate its mean value, this is the forecast for future period. The estimate can be updated as new
information becomes available. The updating techniques are useful when initial estimates are
unreliable or the stability of the average is in question.

Forecasting techniques used for stationary time series data are:

• Naive methods
• Simple averaging methods,
• Moving averages
• Simple exponential smoothing
• Autoregressive moving average (ARMA)

Methods used for time series data with trend are:

• Moving averages
• Holt’s linear exponential smoothing
• Simple regression
• Growth curve
• Exponential models
• Time series decomposition
• Autoregressive integrated moving average (ARIMA)

For time series data with seasonal component, the goal is to estimate seasonal indexes
from historical data. These indexes are used to include seasonality in forecast or remove such
effect from the observed value. Forecasting methods to be considered for these types of data are:

• Winter’s exponential smoothing

• Time series multiple regression
• Autoregressive integrated moving average (ARIMA)

Cyclical time series data show wavelike fluctuation around the trend that tend to repeat.
They are difficult to model because their patterns are not stable. Because of the irregular
behavior of cycles, analyzing these type data requires finding coincidental or leading economic
indicators.

Page 4 of 11
 Forecasting methods to be considered for these types of data are:

• Classical decomposition methods

• Econometric models
• Multiple regression
• Autoregressive integrated moving average (ARIMA)

Qualitative/Subjective Forecasting

In some situations, forecasters rely solely on judgment and opinion to make forecasts. If
management must have a forecast quickly, there may not be enough time to gather and analyze
quantitative data. At other times, especially when political and economic conditions are
changing, available data may be obsolete and more up-to-date information might not yet be
available. Similarly, the introduction of new products and the redesign of existing products or
packaging suffer from the absence of historical data that would be useful in forecasting. In such
instances, forecasts are based on executive opinions, consumer surveys, opinions of the sales
staff, and opinions of experts.

Time Horizon in Forecasting

1. Short-term forecasts (days or weeks) are required for inventory management,

production plans and resource requirements planning.

2. Medium-term forecasts (weeks and months) are required for estimating product
family sales, resource requirements.

3. Long-term forecasts (Months and years) are required for estimating capacity
needs, growth trends.

Qualitative Forecasting

̶ Estimating method that relies on expert human judgment combined with a

rating scale, instead of a hard (measurable and verifiable) data.

Qualitative Approaches

̶ Usually based on judgments about causal factors that underlie the demand of
particular products or services

̶ Do not require a demand history for the product or service, therefore are
useful for new products/services

Page 5 of 11
 ̶ Approaches vary in sophistication from scientifically conducted surveys to
intuitive hunches about future events

̶ The approach/method that is appropriate depends on a product’s life cycle

stage

Qualitative Methods

• Educated guess
• Executive committee consensus
• Delphi method
• Survey of sales force
• Survey of customers
• Historical analogy
• Market research

Qualitative Forecasting Technique Advantages and Problems

Advantages Problems

1. The ability to forecast accurately can be

1. Qualitative forecasting techniques
reduced when forecasters only consider
have the ability to predict changes in
readily available and/or recently perceived
sales patterns.
information.
2. Qualitative forecasting techniques
2. The ability to forecast accurately can be
allow decision makers to incorporate
reduced by the forecasters’ inability to
rich data sources consisting of their
process large amounts of complex
intuition, experience, and expert
information.
judgment.
3. Accurate forecasts can be difficult to
produce when forecasters are overconfident
in their ability to forecast accurately.
4. The ability to accurately forecast may be
significantly reduced by political factors
within organizations, as well as political
factors between organizations.
5. The ability to forecast accurately may be
reduced because of the forecasters’ tendency
to infer relationships or patterns in data when
there are no patterns.

Page 6 of 11
 6. The ability to forecast accurately can be
affected by anchoring; that is, forecasters
may be influenced by initial forecasts (e.g.,
those generated by quantitative methods)
when making qualitative forecasts.
7. Future ability to forecast accurately may be
reduced when a forecaster tries to justify,
rather than understand, a forecast that proves
to be inaccurate.
8. Qualitative forecasting techniques encourage
inconsistencies in judgment due to moods
and/or emotions, as well as the repetitive
decision making inherent in generating
multiple individual product forecasts.
9. Qualitative forecasting techniques are
expensive and time intensive.

Summary: Qualitative Technique Advantages and Problems

Qualitative techniques are a valuable resource for any forecaster. The value of experience
and the ability to analyze complex situations as input to sales forecasts should never be
discounted. Indeed, every sales forecast involves some degree of qualitative input.

Qualitative Techniques in Forecasting

Grass Root

̶ The “Grassroots” technique is based on the concept of asking those who are close to the
eventual consumer, such as salespeople, about what they are going to sell next period,
and adds the forecasts to predict total demand.

̶ Requires an experienced stable work force that knows the customer base.

̶ Can be expensive. Takes sales people away from sales effort. Results can be biased.

̶ In more sophisticated systems, forecasts may be adjusted on the basis of the historical
correlation between the salesperson’s forecasts and the actual sales.

Executive Committee Consensus

̶ Also called Jury of Executive Opinion.

Page 7 of 11
 ̶ Executives from various corporate functions involved in forecasting sales (e.g.,
finance, marketing, sales, production, and logistics) meet to generate forecasts.

̶ Develop medium-long term forecast by asking a group of knowledgeable

executives their opinions with regard to future values of the items being
forecasted.

̶ The presence of a powerful member in the group may prevent reaching

consensus.

Historical Analogy

̶ A judgmental forecasting technique based on identifying a sales history that is

analogous to a present situation, such as the sales history of a similar product, and
using that past pattern to predict future sales.

̶ Existing business can use existing data of the company in the similar situation

̶ New business can use data of other similar business with the same situation.

Marketing Research

̶ Market research is a large and important tool which includes a variety of

techniques, from consumer panels through consumer surveys to test marketing.

̶ The goal of market research is to make predictions about the size/structure of the
market for specific goods and/or services.

̶ These predictions (forecasts) are usually based on small samples and are
qualitative in the sense that the original data typically consist of subjective
evaluations of consumers.

̶ Market research is an important activity in most consumer product firms. It also

plays an increasingly important role in the political and electoral process.

Delphi Method

̶ Iterative group process allows experts to make forecasts without meeting face-to-
face

̶ Participants are typically 5-10 experts who make the forecast

Page 8 of 11
 ̶ Staff personnel assist by preparing, distributing, collecting, and summarizing a
series of questionnaires and survey results

̶ After three or four passes through this process, a consensus forecast typically
emerges.

When the Delphi method is used for forecasting, the input of experts, either internal or
external to a company, is solicited and proceeds as follows:

1. Each member of the panel of experts who is chosen to participate writes an answer to the
question being investigated (e.g., a forecast for product or industry sales) and all the
reasoning behind this forecast.

2. The answers of the panel are summarized and returned to the members of the panel, but
without the identification of which expert came up with each forecast.

3. After reading the summary of replies, each member of the panel either maintains his or
her forecast or reevaluates the initial forecast and submits the new forecast (and the
reasoning behind changing his or her forecast) in writing.

The answers are summarized and returned to panel members as many times as necessary
to narrow the range of forecast. An appropriate use of the Delphi method is for the prediction of
mid- to long-term company sales levels or long-term industry sales levels.

When this technique is used within a company, it can be thought of as a kind of “virtual”
jury of executive opinion, because the executives do not meet face to face. The purpose of this
distance is to allow each member to use his or her reasoning to develop a forecast, without the
influence of strong personalities or the fact that the “boss” has a pet forecast.

The Delphi method also reduces the effects of groupthink on the decision-making
process. Since the participants do not meet face to face, the bias that occurs because of a desire
on the part of group members to support each other’s positions or the influence of a strong leader
within the group is minimized.

Problems with this method of qualitative forecasting focus on its tendency to be

unreliable, that is, the outcomes can be highly dependent on the composition and expertise of
panel members. To some extent this source of bias is the result of group members not being
willing/able to seek out information other than what is readily available and/or recently
perceived.

Decision Analysis Tools for Qualitative Forecasting

Page 9 of 11
 ̶ Decision Tree Diagrams - The advantage of decision tree diagrams is that they enable
participants to visualize the context of a complex decision, thereby reducing biases that
occur because of limitations on forecasters’ abilities to process complex information.

̶ Simulations - Another decision analysis tool that can be used in the qualitative
forecasting decision process is simulation. Like decision tree diagrams, simulation
requires forecasters to structure their decision making by examining and stating their
assumptions. Essentially, simulation demands that the system under investigation be
defined, enabling that system to be manipulated so that “what-if” analyses can be
performed to explore alternatives

Some Short-Range Forecasting Methods

Naïve Forecast

A simple but widely used approach to forecasting is the naive approach. A naive forecast
uses a single previous value of a time series as the basis of a forecast. The naive approach can be
used with a stable series (variations around an average), with seasonal variations, or with trend.
With a stable series, the last data point becomes the forecast for the next period. Thus, if demand
for a product last week was 20 units, the forecast for this week is 20 units.

With seasonal variations, the forecast for this “season” is equal to the value of the series
last “season.” For example, the forecast for demand for Christmas trees this Christmas season is
equal to demand for Christmas tress last Christmas.

Example:

Period Actual Forecast

1 50
2 53 50
3 52 53
4 52

Moving Average

A moving average forecast uses a number of the most recent actual data values in
generating a forecast. In moving average forecast, you get the average of the most recent
forecasts and use it as the forecast for the next period. The number of entries to use in averaging
depends on the forecaster. You may use 2-periods, 3-periods, 4-periods, etc.

Example:

In this example, we will be using 3-periods in forecasting. The first 3 data in the demand
will be used for the 4th period forecast. As new actual data is recorded, the 3 most recent data will
now be used for the 5th period data and so on.

Page 10 of 11
 Period Demand Forecast
1 42
2 40 ( 42+ 40+ 43 ) ÷ 3=41.67
3 43
4 40 41.67
5 41 41.00
6 38 41.33
7 39.67

( 40 +43+ 40 ) ÷3=41.00

Weighted Simple Average

A weighted average is similar to a moving average, except that it assigns more weight to
the most recent values in a time series. For instance, the most recent value might be assigned a
weight of .40, the next most recent value a weight of .30, the next after that a weight of .20, and
the next after that a weight of .10. Note that the weights must sum to 1.00, and that the heaviest
weights are assigned to the most recent values. The weights assigned will depend on the
forecaster as long as the total weights is equal to 1.

Example:

In this example we will use 0.40, 0.30, 0.20 and 0.10 as weights. Same with the previous
method, moving average, the most recent forecasts is used to forecast for the next period. As new
data is recorded, it will be used to compute for the next period’s forecast.

Period Demand Forecast

1 42
( 42 × 0.10 ) + ( 40 × 0.20 ) +
2 ( 43 × 0.30 ) + ( 40 × 0.40 )=41.1040
3 43
4 40
5 41 41.10
6 39 41.00
7 40.20

( 40 × 0.10 ) + ( 43× 0.20 ) + ( 40× 0.30 ) +(41 ×0.40)=41.00

Page 11 of 11