manufacture, number of employees engaged in certain kinds of  Survey of customers buying plan

manufacturing, number of manufacturing establishments, Quantitave Methods

person-hours worked, total value of shipments of particular
 Projection of Past Sales
items, and capital expenditures for new plant and equipment.
 Time series Analysis – Moving Average Method
    
 Exponential smoothening
Sales potentials, are defined earlier, are quantitative estimates of
the maximum possible sales opportunities present in particular  Regression Analysis
market segments open to a specified company selling a good or  Econometric model building and simulation
service during a stated future period. They are derived from
market potentials after analyses of historical market share
relationships and adjustments for changes in companies’ and Forecasting Methods

competitors’ selling strategies and practices.

A firm’s sales potential and its sales forecast are not usually
identical – in most instances, the sales potential is larger than Qualitative Quantitative
the sales forecast. There are several reasons for this: Methods Methods
“ Some companies do not have sufficient production capacity
• Projection of Past Sales
to capitalize on the full sales potential; • Jury of Executive opinion
• Time series Analysis – Moving
•The Delphi Technique
“ Other firms have not yet developed distributive networks •Poll of Sales force opinion
Average Method
• Exponential smoothening
capable of reaching every potential customer; •Survey of customers
•Regression Analysis
buying plan
“ Others do not attempt to realize their total sales potentials • Econometric model building
because of limited financial resources; and
“ Still others, being more profit oriented than sales oriented, Let us study Qualitative Methods in this lesson .Quantitative
seek to maximize profitable sales and not possible sales. Methods are discussed in Next Lesson
The estimate for sales potential indicates how much a company   
could sell if it had all the necessary resources and desired to use

them. The sales forecast is a related but different estimate – it
indicates how much a company with a given amount of There are two steps in this method:
resources can sell if it implements a particular marketing 1. high-ranking executives estimate probable sales, and
program. 2. an average estimate is calculated.
   The assumption is that the executives are well informed about
A sales forecasting method is a procedure for estimating how the industry outlook and the company’s market position,
much of a given product (or product line) can be sold if a given capabilities, and marketing program. All should support their
marketing program is implemented. No sales forecasting estimates with factual material and explain their rationales.
method is foolproof – each is subject to some error. Some Companies using the jury of executive opinion method do so
methods are unsophisticated, such as the jury of executive for one or more of four reasons:
opinion or the poll of sales force opinion. Others involve the
1. This is a quick and easy way to turn out a forecast.
application of sophisticated statistical techniques, such as
regression analysis or econometric model building and simula- 2. This is a way to pool the experience and judgment of well-
tion. Two sales forecasting methods may be either sophisticated informed people.
or unsophisticated, depending upon how they are used – the 3. This may be the only feasible approach if the company is
projection of past sales and the survey of customers’ buying so young that it has not yet accumulated the experience to
plans. use other forecasting methods.
Well-managed companies do not rely upon a single sales 4. This method may be used when adequate sales and market
forecasting method but use several. If different methods statistics are missing, or when these figures have not yet
produce roughly the same sales forecasts, then more confidence been put into the form required for more sophisticated
is placed in the results. But if different methods produce greatly forecasting methods.
different sales forecasts, then the sales situation merits further The jury of executive opinion method has weakness.
study.
• Its findings are based primarily on opinion, and factual
The forecasting methods can be broadly classified as qualitative evidence to support the forecast is often sketchy.
and quantitative .
• This approach adds to the work load of key executives,
Qualitative methods are: requiring them to spend time that they would otherwise
 Jury of Executive opinion devote to their areas of main responsibility.
 The Delphi Technique • And a forecast made by this method is difficult to break
 Poll of Sales force opinion down into estimates of probable sales by products, by time
intervals, by markets, by customers, and so on.


 method of getting an alternative estimate for use as a check on a
Several years ago researchers at the he Rand Corporation sales forecast obtained through some other approach.
developed a technique for predicting the future that is called the

Delphi technique. This is a version of the jury of executive
What more sensible way to forecast than to ask customers
opinion method in which those giving opinions are selected for
about their future buying plans? Industrial marketers use this
their “expertise”. The panel of experts responds to a sequence
approach more than consumer-goods marketers, because it is
of questionnaires in which the responses to one questionnaire
easiest to use where the potential market consists of small
are used to produce the next questionnaire. Thus, information
numbers of customers and prospects, substantial sales are
available to some and not to other experts is disseminated to
made to individual accounts, the manufacturer sells direct to
all, enabling all to base their final forecasts on “all available”
users, and customers are concentrated in a few geographical areas
information. Some contend that “this technique eliminates the
(all more typical of industrial than consumer marketing). In
bandwagon effect of majority opinion.”
such instances, it is relatively inexpensive to survey a sample of
 customers and prospects to obtain their estimated requirements
In the poll of sales force opinion method, often tagged “the for the product, and to project the results to obtain a sales
grass-roots approach,” individual sales personnel forecast sales forecast. Survey results, however, need tempering by manage-
for their territories; then individual forecasts are combined and ment’ specialized knowledge and by contemplated changes in
modified, as management thinks necessary, to form the marketing programs. Few companies base forecasts exclusively
company sales forecast. on a survey of customers’ buying plans. The main reason lies in
This approach appeals to practical sales managers because : the inherent assumptions that customers know what they are
going to do and that buyers’ plans, once made, will not change.
• forecasting responsibility is assigned to those who produce
the results. Even though the survey of customers’ buying plans is generally
an unsophisticated forecasting method, it can be rather sophisti-
• Furthermore, there is merit in utilizing the specialized
cated – that is, if it is a true survey (in the marketing research
knowledge of those in closet touch with market
sense) and if the selection of respondents is by probability
conditions. Because the salespeople help to develop the
sampling. However, since it gathers opinions rather than
forecast, they should have greater confidence in quotas
measures actions, substantial nonsampling error is present.
based upon it.
Respondents do not always have well-formulated buying plans,
• Another attractive feature is that forecasts developed by this and, even if they do, they are not always willing to relate them.
method are easy to break down according to products, In practice, most companies using this approach appear to pay
territories, customers, middlemen, and sales force. little attention to the composition of the sample and devote
But the poll of sales force opinion approach has weaknesses. minimum effort to measuring sampling and nonsampling
• Not generally trained to do forecasting, and influenced by errors.
current business conditions in their territories, salespersons 
tend to be overly optimistic or overly pessimistic about In this lesson we studied the about the personal selling
sales prospects. objectives which can be qualitative and quantitative. Then we
• They are too near the trees to see the forest — they often discussed about market potential which is sales opportunities
are unaware of broad changes taking place in the economy present for all sellers of a good or service. Sales potential which
and of trends in business conditions outside their own is the maximum sales opportunity for a company selling a good
territories. or service. We have also discussed the qualitative forecasting
• Furthermore, if the “forecasts” of the sales staff are used methods.
in setting quotas, some sales personnel deliberately
underestimate so that quotas are reached more easily.
• To some extent, the weakness of this method can be
overcome through training the sales force in forecasting
techniques, by orienting them to factors influencing
company sales, and by adjusting for consistent biases in
individual salespersons’ forecasts.
For most companies, however, implementing corrective actions
is an endless task, because sales personnel turnover is constantly
going on, and new staff members (whose biases are unknown
at the start) submit their forecasts along with those of veteran
sales personnel with known biases. In short, this method is
based to such a large extent on judgment that it is not appropri-
ate for most companies to use it as the only forecasting
method. The poll of sales force opinion serves best as a


 


  One drawback of time-series analysis is that it is difficult to “call

To learn about the quantitative sales forecasting methods the turns”. Trend and cycle analysis helps in explaining why a
trend, once under way, continues, but predicting the turns often
• Projection of Past Sales
is more important. When turns for the better are called correctly,
• Time-series analysis management can capitalize upon sales opportunities; when
• Moving Average method turns for the worst are called correctly, management can cut
• Exponential smoothing losses.
• Regression Analysis   
• Econometric Model Building and Simulation Are used to allow for market place factor changing at different
rates and at different times . With this method both distant
   past and distant future have little value in forecasting . The
In This lesson we will learn about Quantitative forecasting moving average is a technique that attempts to “smooth out “
Methods. Let us discuss each one by one.
The different rates of change for the immediate past , usually
    past three to five years. The forecast is the mean of these past
The projection of past sales method of sales forecasting takes a periods and is only valid for one period in future. The forecast
variety to forms. The simplest is to set the sales forecast for the is updated by eliminating the data for the earliest period and
coming year at the same figure as the current year’s actual sales, adding the most recent data.
or the forecast may be made by adding a set percentage to last Take for Example in the following table 7.1 .The sales volumes
year’s sales, or to a moving average of the sales figure for several for periods 3 , 4, 5 are totaled and divided by 3 to derive the
past years. For instance, if it is assumed that there will be the mean of 366.6 which is the period 6 forecast.If the company
same percentage sales increase next year as this year, the fore- operates in the stable environment a short 2 or 3 years average
caster might utilize a naïve model projection such as will be most useful . For a firm in an industry with cyclical
this year’s sales variation , the moving average should use data equal to length
next year’s sales = this year’s sales x
last year’s sales of cycle.
This year’s sales are inevitably related to last year’s. Similarly, next
year’s sales are related to this year’s and to those of all preceding Table 7.1
years. Period Sales volume Sales for three Three year moving
Projecting present sales levels is a simple and inexpensive Year period Average
forecasting method and may be appropriate for companies in 1 200
more or less stable or “mature” industries — it is rare in such 2 250
industries for a company’s sales to vary more than 15 percent 3 300 750
plus or minus from the preceding year. 4 350 900 300
 5 450 1100 / 3 = 366.6
Not greatly different in principle from the simple projection of 6 ?
past sales in time-series analysis, a statistical procedure for Period 6 forecast = 366.6
studying historical sales data. This procedure involves isolating
and measuring four chief types of sales variations: long-term 
trends, cyclical changes, seasonal variations, and irregular One statistical technique for short-range sales forecasting,
fluctuations. Then a mathematical model describing the past exponential smoothing, is a type of moving average that
behavior of the series is selected, assumed values for each type represents a weighted sum of all past numbers in a time series,
of sales variation are inserted, and the sales forecast is ‘cranked with the heaviest weight placed on the most recent data. To
out.” illustrate, consider this simple but widely used form of
For most companies, time-series analysis finds practical exponential smoothing —— a weighted average of this year’s
application mainly in making long-range forecasts. Predictions sales is combined with the forecast of this year’s sales to arrive at
on a year-to-year basis, such as are necessary for an operating the forecast for next year’s sales. The forecasting equation, in
sales forecast, generally are little more than approximations. other words, is
Only where sales patterns are clearly defined and relatively stable next year’s sales = a (this year’s sales) + (1-a)(this year’s forecast)
from year to year is time-series analysis appropriately used for
The a in the equation is called the “smoothing constant” and is
short-term operating sales forecasts.
set between 0.0 and 1.0. If, for example, actual sales for this year


came to 320 units of product, the sales forecast for this year was product and its own sales, but the correlation coefficient was
350 units, and the smoothing constant was 0.3, the forecast for too low to use in forecasting company sales. The same was true
next year’s sales is of personal disposable income and retail sales; their correlation
(0.30)(320) + (0.7)(350) = 341 units of products coefficients with company sales were too low to use in forecast-
ing company sales.
Determining the value of a is the main problem. If the series
of sales data changes slowly, a should be small to retain the
effect of earlier observations. If the series changes rapidly, a
should be large so that the forecasts respond to these changes.
In practice, a is estimated by trying several values and making
retrospective tests of the associated forecast error is then chosen
for future smoothing.
Evaluation of past sales projection methods. The key limitation
of all past sales projection methods lies in the assumption that
past sales history is the sole factor influencing future sales. No
allowance is made for significant changes made by the company
in its marketing program or by its competitors in theirs. Nor is
allowance made for sharp and rapid upswings or downturns in
business activity, nor is it usual to correct for poor sales
performance extending over previous periods.
The accuracy of the forecast arrived at through projecting past
sales depends largely upon how close the company is to the
market saturation point. If the market is nearly 100 percent
saturated, some argue that it is defensible to predict sales by
applying certain percentage figure to “cumulative past sales of
the product still in the hands of users” to determine annual Automobile sales (millions of units)
replacement demand. However, most often the company whose The tyre manufacturer measured the relationship between its
product has achieved nearly 100 percent market saturation finds, own dollar sales and unit sales of automobiles and found a
since most companies of this sort market durables or semi much higher degree of correlation . The dots on this scatter
durables, that its prospective customers can postpone or diagram cluster closely around the straight line that is the result
accelerate their purchases to a considerable degree. of the mathematical computation between the two series of
Past sales projection methods are most appropriately used for data. If the correlation had been perfect, all the dots would have
obtaining “check” forecasts against which forecasts secured fallen on the line.
through other means are compared. Most companies make Where sales are influenced by two or more independent
some use of past sales projections in their sales forecasting variables acting together multiple regression techniques are used.
procedures. The availability of numerous computer programs Evaluation of regression analysis for sales forecasting.
for time-series analysis and exponential smoothing has If high coefficients of correlation exist between company sales
accelerated this practice. and independent variables, the forecasting problem is simpli-
  fied, especially if the variables “lead” company sales. The
Regression analysis is a statistical process and, as used in sales probable course of sales may then be charted, and the forecaster
forecasting, determines and measures the association between can concentrate on factors that might cause deviations. But it is
company sales and other variables. It involves fitting an necessary to examine other circumstances that might upset past
equation to explain sales fluctuations in terms of related and relationships. A forecast made through regression analysis
presumably causal variables, substituting for these variables assumes that past relations will continue. A “lead-lag” associa-
values considered likely during the period to be forecasted, and tion in which deviations regularly occur in related independent
solving for sales. In other words, there are three major steps in variable(s) prior to a change in company sales is a near-ideal
forecasting sales through regression analysis: situation, but it rarely holds except over short periods. Lead-lag
1. Identify variables causally related to company sales. relationships are common, but associations between the lead
variables and sales in which the intervening time intervals
2. Determine or estimate the values of these variables related
remain stable are uncommon. Periods not only contract or
to sales.
expand erratically; they vary greatly during different phases of
3. Derive the sales forecast from these estimates. the business cycle.
Computers make it easy to use regression analysis for sales If close associations exist between company sales and a reliable
forecasting. One tyre manufacturer, for instance, used simple barometer, estimates are improved by experts’ predictions of
regression analysis to determine the association between probable changes in the barometer. However, one danger in
economic variables and its own sales. This company discovered using regression analysis is that forecasters may put too much
that a positive correlation existed between gross national faith in the statistical output. They may abandon independent


appraisals of future events because of a statistically developed in service, and ten years later, 54. For this batch of 10,000
forecast. It is wrong to place blind faith in any forecasting products units, scrappage is 1,035 in the fifth year (that is, 1,379-
method. It is wise to check results with those of other forecasts. 344, the difference between the accumulated total scrappage at
the close of the fifth and fourth years, respectively). In the fifth
   
year, then, 1,035 replacement sales trace back to the batch of

10,000 product units that entered service five years before.
Econometric model building and simulation is attractive as a
sales forecasting method for companies marketing durable New-owner demand is the net addition to users’ stocks of the
goods. This approach uses an equation or system of equations product that occurs during a given period. For instance, if
to represent a set of relationships among sales and different 2,000,000 units of some appliance were in service at the start of
demand-determining independent variables. Then, by “plug- a period and 2,500,000 at the end, new-owner demand was
ging in” values (or estimates) for each independent variable 500,000 during the period. Forecasting the number of sales to
(that is, by “simulating” the total situation), sales are forecast. new owners involves treating the stock of the product in the
An econometric model (unlike a regression model) is based hands of users as a “population” exhibiting “birth” and
upon an underlying theory about relationships among a set of “death” characteristics, that is, thinking of it as being analogous
variables, and parameters are estimated by statistical analysis of to a human population.
past data. An econometric sales forecasting model is an Constructing this sort of econometric model requires going
abstraction of a real-world situation, expressed in equation through three steps. First, study independent variables affecting
form and used to predict sales. For example, the sales equation Figure 7.2 Durable-Goods Survival Coefficients (Maxi-
for a durable good can be written. mum Service
S = R + N Year Survival/Coefficient
Where 1. 0000
S = total sales 0. 9995
R = replacement demand (purchases made to replace product 0. 9946
units going out of use, as measured by the scrappage of
old units) 0. 9656

N = new-owner demand (purchases made not to replace 0. 8621

existing product units, but to add to the total stock of the 0. 6406
product in users’ possession) 0. 3594
Total sales of a durable good, in other words, consist of 0. 1379
purchases made to replace units that have been scrapped and 0. 0344
purchases by new owners. Thus, a family that has a five-year-old
machine trades it in to a dealer as part payment for a new 0. 0054
machine and becomes part of the replacement demand 0. 0000
(although only effectively so when the five-year-old machine,
perhaps passing through several families’ hands in the process, each demand category (replacement and new owner) and choose
finally comes to be owned by a family that goes ahead and for correlation analysis those that bear some logical relationship
consigns its even-older machine to the scrap heap). to sales (the dependent variable). Second, detect that combina-
Replacement demand is measured by the scrappage of old units tion of independent variables that correlates best with sales.
of products, that it, by the percentage of the total stock of the Third, choose a suitable mathematical expression to show the
product in users’ hands that is taken out of service through quantitative relationships among the independent variable and
consignment to the trash pile, by sale to a junk dealer, or merely sales, the dependent variables. This expression becomes the
by being stowed away and never used again. Replacement econometric model.
demand in any one year does not include demand originating The procedure for building econometric models is simple, but
from the family that had a five-year-old machine that it traded finished models can take on formidable appearances. Consider,
to a dealer for a new machine, with the dealer reselling the old for example, this econometric model for forecasting sales of
machine to another family who buys it second-hand. Only washing machines.
when a particular machine goes completely out of service is it
regarded as scrapped, and, at that time (through a chain of { [
Stc = Yt – yt+ Yt Ht (0.03 – 0.0157) ((It + 3Ct)/P1 )] - 0.0000283Y1 }
purchases and trade-ins), some family becomes a part of
replacement demand. Econometricians estimate replacement (100.01818t – 33.1143 )
demand by using life expectancy of survival tables, which are where
similar to the life (or mortality) tables used by life insurance Stc = calculated value for forecasted sales of washing machines
actuaries. An example is shown in Figure 7.2 during some time period
If some durable good has a maximum service life of eleven Yt = level of consumers’ stock that would occur in the
years and 10,000 units of the good enter service in some year, following period
the table indicates that five years later, 8,621 will probably still be


 (as of January 1) if no washing machines were sold and weaknesses (as well as marketing programs) against those of
scrappage rates competitors. The result is an estimate of expected market share
yt = level of consumers’ stock that would occur in the that (when applied to the industry sales forecast) results in a
following forecast of company sales. The poll of sales forece opinion
method leaves this appraisal up to the sales personnel – they
period (as of January 1) if no washing machines were sold
focus on estimating how much the company can sell, not on
and scrappage rates remained the same
how much the industry can sell. Unsophisticated forms of the
Ht = number of wired (i.e., electrified) dwelling units, in past sales projection method implicitly assume that no changes
millions will occur in the company’s strengths and weaknesses not in its
It = disposable personal income marketing programs vis-a-vis those of its competitors. In the
Ct = net credit extended (excluding credit extended for other four forecasting methods considered in this chapter,
automobiles) management makes this appraisal when it determines the
company’s probable market share percentage. Moreover,
Pt = price index for house furnishings
although some companies check such appraisals with sales
100.01818t – 33.1143 = trend of real purchasing power over time personnel, in most the main appraisal of competitive position
It + 3Ct/Pt = real purchasing power is made by executives better informed on the overall sales
Thus, new owner demand in this model is represented by Yt - outlook than any sales-person can be.
yt, determined by applying appropriated survival coefficients to Forecasting a company’s market share varies in complexity from
previous years’ sales of washing machine and estimating one industry to another. In the steel industry, the number of
consumers’ total stocks of washing machines in each year. competitors is small and market share is stable, so determining
Replacement demand is represented by the other symbols in the a given company’s market share is a simple task – a matter of
equation and takes into account the number of wired dwelling projection past trends and adjusting for anticipated changes in
units (washing machines are not sold to people who live in the company’s relative strengths and weaknesses. But in the
homes with no electricity) real purchasing power (disposable women’s clothing industry, the number of competitors is large
personal income plus credit availability divided by a price index), and market shares fluctuate widely, so determination of market
and real purchasing power adjusted for the historical trend. share is difficult. The ability to evaluate a clothing style’s
Regression analysis was used to derive the numerical values in salability is a key element in forecasting, and this requires both
this model. thorough knowledge of market trends and keen judgment.
Econometric model building seems a nearly ideal way to Most companies operate in industries that lie somewhere
forecast sales. Not only does it consider the interaction of between these two extremes, with market shares neither as
independent variables that bear logical and measurable relation- stable as in steel nor as volatile as in women’s apparel. Forecast-
ships to sales, it uses regression analysis techniques to quantify ers in most companies need information on competitors’ plans
these relationships. Econometric models, however, are best to launch new and improved products, advertising and selling
used to forecast industry sales not the sales of individual plans, pricing strategies, and so on. When forecasters evaluate
companies. This is because the independent variables affecting this information in relation to their own company’s proposed
an individual company’s sales are more numerous and more marketing and selling plans, they are in a position to exercise
difficult to measure than are those determining the sales of an informed judgment in predicting the company’s probable
entire industry. Many companies use an econometric model to market share. If, for example, a forecaster knows that a major
forecast industry sales, and then apply an estimate of the competitor plans a substantial price cut on a product that many
company’s share-of the-market percentage to the industry buyers buy mainly on the basis of price, it will be necessary to
forecast to arrive at a first approximation for the company’s lower the estimate of the company’s market share unless
forecasted sales. management is willing to match the price cut. Forecasting a
company’s market share is a matter both of examining past
    trends and of appraising impending changes in competitive
   relationships.
Many companies forecast both their own sales and sales of the
industry. Of those using multimethod forecasting procedures,
nearly all – at one or more stages – provide for the making of
an industry sales forecast. In fact, of the six sales forecasting
methods discussed in this chapter, only in two – the poll of
sales force opinion and unsophisticated forms of projecting
past sales – is to normal to skip the industry sales forecast and
forecast company sales directly. The general practice is to forecast
industry sales early in the procedure and from it derive a
company sales forecast for use as a check against forecasts arrived
at through other methods.
Deriving a company sales forecast from an industry sales
forecast requires an appraisal of company strengths and

