DEMAND FORECASTING

The activity of estimating the quantity of a product or service that

consumers will purchase. Demand forecasting involves
techniques including both informal methods, such as educated
guesses, and quantitative methods, such as the use of historical
sales data or current data from test markets. Demand forecasting
may be used in making pricing decisions, in assessing future
capacity requirements, or in making decisions on whether to
enter a new market.

— according to Cundiff and Still, "Demand Forecasting is an

estimate of Demand during a specified period. Which estimate is
tied to a proposed marketing plan and which assumes a particular
set of uncontrollable and competitive forces."

— In the words of Prof. Philip Kotler. The company (sales)

forecast is the expected level of company sales based on a chosen
marketing plan and assumed marketing environment"

— According to Evan J. Douglas, "Demand forecasting may be

defined as the process of finding values for demand in future time
periods."

Demand Forecasting: The cost should be controlled by

producing correct level of goods in the firm and also according to
the demand for those goods in the market. For the estimation of
demand, demand forecasting is to be done by the firm.

 Forecasting = estimation of future situations.

 Forecasting reduces or minimizes the uncertainty.
 By forecasting effective decisions can be taken for
tomorrow.
 Demand forecasting is based on the determinants of the
demand.
 Demand for goods increases and gives sales.
 Sales are the primary source of the income for a firm.
STEPS INVOLVED IN DEMAND FORECASTING

1. Identification of business objectives: In the first stage we

should know what is the aim of forecasting? What we get or know
from the forecasting? Estimation of factors like quantity and
composition of demand for goods, price to be quoted, sales
planning and inventory control etc., are done in the first stage.

2. Determining the nature of goods under consideration:

Different category of goods has their own distinctive
demand. Example capital goods, consumer durables and non-
durables goods in which category our goods fall we should
estimate.

3. Selecting a proper method of forecasting: There are

different methods for demand forecasting. Which is best suited
method that we should select for doing demand forecasting?

4. Interpretation of results:The forecasting which is done by

the managerial economist should be interpreted in detailed
manner. That means it should be easy to understand by the top
management.
Demand Forecasting Techniques
To invest money and others factors in business; we require a
reasonable accurate forecast of demand. Starting with qualitative
methods like survey of collective opinions, buyers' intention,
Delphi approach and its variant, a number of quantitative
methods are used for computing demand forecasts as detailed
below:

Opinion polling method

a) Collective opinion Survey: Sales personnel are closest

to the customers and have an intimate feel of the market.
Thus they are most suited to assess consumer’s reaction to
company's products. Here each salesperson makes an
estimate of the expected sales in their area, territory, state
and/or region, These estimates are collated, reviewed and
revised. Taking in to account product design, features and
price is decided and made. Thus, "collective opinion survey
forms the basis of market Analysis and demand forecasting.
Although this method is simple, direct, first hand and most
acceptable, it suffers from following weaknesses

1. demand estimates by individual salespersons to obtain total

demand of the country may be risky as each person has
knowledge about a small portion of market only
2. Salesperson may not prepare the demand estimation with
the seriousness and care
3. limited experience in their employment, salesperson may
not have the required knowledge and experience
b) Survey of Customers Intention: Another method of
demand forecasting is to carry out a survey of what
consumers prefer and intend to buy. If the product is sold to
a few large industrial buyers, survey would involve
interviewing them. If it is a consumer durable product, a
sample survey is carried out about what they are planning or
intending to buy. It is not east to query all consumers
 through direct contact or through printed questionnaire by
mail. These surveys serve useful purpose in establishing
relationships between
a) demand and price
b) demand and income of consumers
c) demand and expenditure on advertisement etc.

This method is preferred when bulk of the sales made to

institutions and industrial buyers and only a few of them
have to be contacted. Disadvantages are. Survey method is
not useful for households - interviewing them is not only
difficult but also expensive. They are not able to give precise
idea about their intentions particularly when alternative
products are available in the market.

c) Delphi Method: The Delphi technique was developed

at RAND Corporation in the 1950s. Delphi method is a group
(members) process and aims at achieving a `single opinion
of the members on the subject. Herein experts in the field of
marketing research and demand forecasting are engaged
in

 analyzing economic conditions

 carrying out sample surveys of market
 conducting opinion polls

Based on the above, demand forecast is worked out in

following steps:
1. Administrator sends out a set of questions in writing
to all the experts on the panel, who are requested to write
back a brief predication.
2. Written predictions of experts are collected and
combined, edited and summarized together by the
administrator.
3. Based on the summary, administrator designs a new
set of questions and gives them to the same experts who
answer back again in writing.
 4. Administrator repeats the process of collecting,
combining, editing and summarizing the responses.
5. Steps 3 and 4 are repeated by the administrator to
experts with diverse backgrounds until they come to one
single opinion.

If there is divergence of opinions and hence conclusions,

administrator has to sort it out through mutual discussions.
Administrator has to have the necessary experience and
background as he plays a key role in designing structured
'questionnaires and synthesizing the data.

d) Nominal Group Technique: This technique was

originally developed by Delbecq and VandeVen. This is a
further modification of Delphi method of forecasting. A panel
of 3-4 groups of up to 10 experts are formed and allowed to
interact, discuss 'and rank all the suggestions in descending
(highest to lowest) order as per the following procedure:

Experts sit around a table in full view of one another

and are asked to speak to each other. An administrator hand
over copies of questionnaire needing a forecast and each
expert is expected to write down a list of ideas about the
questions. After everyone has written down their ideas,
administrator asks each expert to share one idea, out of own
list. The idea shared is written on the `flip chart' which
everyone can see. Experts give ideas in rotation until all of
them are written on the `flip chart'. No discussion takes
place in this phase and usually 15 to 25 ideas emerge from
this format. In the next phase, experts discuss ideas
presented by them. Administrator ensures that all ideas
have been adequately discussed. During discussions similar
ideas are combined. This reduces the number of ideas. After
completing group discussions, experts are asked to give in
writing ranks to ideas according to their perception of
priority
Statistical methods

 Trend projection method

This technique assumes that whatever past years demand

pattern will be continued in the future also. Basing on the
historical data that means previous year’s data is used to
predict the demand for the future. In this trend projection
method, previous year’s data is presented on the graph and
future demand is estimated.

 Regression Analysis

Past data is used to establish a functional relationship

between two variables. For Example, demand for consumer
 goods has a relationship with income of Individuals and
family; demand for tractors is linked to the agriculture
income and demand for cement, bricks etc. are dependent
upon value of construction contracts at any time.
Forecasters collect data and build relationship through co-
relation and regression analysis of variables.

How to forecast demand for new products?

Demand forecasting for the new products requires special skill
and techniques as they are new products and no previous data
will be available about their sales.

The method or techniques should be carefully tailored for the

product. Joel Dean makes six possible approaches towards
forecasting of new products. They are as follows:
1. The Evolutionary approach in forecasting demand
The principle behind this approach is that the demand for a new
product is only an outgrowth and evolution of the existing
product. It means that the demand conditions of the existing
product should be taken into account while accessing the demand
for the product.

Examples: Colour TV sets from black and white TV sets; Left-side

steering cars from right-side steering cars, etc. But this approach
is useful only when the new product is very close to the old
existing product.

2. Substitute approach in forecasting demand

By this the new product is analyzed as a substitute for the old
existing product or service.

3. Growth curve approach in forecasting demand

The estimates of rate of growth and ultimate level of demand for
the new product will be established on the basis of some growth
patterns of an already established product.
For example, the average sales of Talcum powder will give an idea
as to how a new cosmetic will be received in the market.

4. Opinion Poll approach in forecasting demand

Under this, the demand for the new product will be estimated by
making direct enquiries from the ultimate consumers. This is
done by sample survey method. But, this is a very complicated
process as there will be problems of sampling, probing the real
intentions of the consumers, etc..

5. Sales Experience approach in forecasting demand

According to Sales experience approach method, samples of new
products shall be offered in a sample market to forecast demand.
This is done through distributive channels like departmental
stores or cooperative society, etc., or by direct mailing. Total
demand is predicted on the basis of the sample market. But, the
difficulty in this lies in determining the allowance to make for the
immaturity of the sample market and full-fledged market.

6. Vicarious approach in forecasting demand

Through vicarious approach method, the reaction of the customer
towards new product can be found out indirectly through the
specialized dealers who are able to judge the needs, tastes and
preferences of customers.

The dealers being the link between the producer and the ultimate
consumers, will be able to know how the customers will receive
the new product.

10 steps for forecasting demand and revenues for new

products
Forecasting demand and revenues for new variants of existing
products is difficult enough. But forecasting for radically
innovative products in emerging new categories is an entirely
different ball game. There are no past trends to reassuringly
extrapolate into the future, just a ton of uncertainty about
whether the latent demand that the marketing folk suggested to
secure the R&D funding is real or not. And after so much
investment, the board is sure that this is the product that is going
to become the next cash cow. Sure, you could manage their
expectations by reminding them that something like 80% of new
products fail and name drop a few of the spectacular flops of
Fortune 500 companies. But that would be career limiting. A
better alternative is to take control of the situation and adopt
some of the forecasting best practices approaches that others
have found to work.
Step 1: Make it a collaborative effort
Identify a handful of key people from marketing, sales, operations,
and relevant technical departments and form a working group.
This core team will be responsible for developing and managing
the reforecasting process through the launch period until demand
planning becomes more predictable.
Step 2: Identify and agree upon the assumptions
Collectively review all the available qualitative and quantitative
data from market research, market testing, and buyer surveys.
Use the data to identify a set of assumptions that can form the
basis of a forecasting model. Ideally this will include assumptions
about:

 Number of consumers in the target market

 Proportion expected to buy the product
 Anticipated timing of their purchase
 Patterns of repeat purchasing and replacement purchasing
Be prepared to commission additional research or consult
external industry experts to fill any important data gaps. And
always let the working group use their collective judgement to
identify a realistic range of values for each assumption.

Step 3: Build granular models

Not all consumers will purchase a new product at the same rate.
Some may be prepared to queue all night around the block to get
their hands on it, but others will want to wait for subsequent
versions when any unforeseen bugs are fixed and prices are
typically lower. So it is important to build a forecasting model that
is sufficiently granular to reflect how and when different market
segments in different geographies might purchase the product
and at what price.

Step 4: Use flexible time periods

Sales over the first few days and weeks in the life of any new
product need to be carefully monitored as they will quickly show
how demand is likely to grow in the future. So although the sales
and finance function may only be interested in monthly data, it
pays to develop detailed daily forecasts for the first quarter
against which to track actual sales.

Step 5: Generate a range of forecasts

Run through a number of iterations, changing various
assumptions and probabilities in the model to generate a range of
forecasts. This is easily done if a modelling solution that can be
recalculated in real-time is deployed as internal experts and
business leaders can generate and test alternative scenarios on
the fly.

Step 6: Deliver the outputs that users need quickly

In planning for a new product launch, agreements may have been
reached with a number of suppliers to deliver rapid
replenishment designed to prevent stock outs in the most
uncertain period immediately after the launch. However if
reforecasting the exact replenishment needs of every distribution
point in the supply chain involves multiple steps, much of that
precious time will evaporate.

Building a fully integrated forecasting model that compares

existing stock level and automatically generates a detailed
replenishment report for every location as soon as any high level
assumptions change precludes such delays and shortens the
replenishment cycle.
Step 7: Combine different techniques
Bottom up modelling based on purchasing intentions is not the
only method available for forecasting demand for new products.
In some markets, such as technology and consumer electronics,
products can go through an entire life cycle in a matter of months.
Such narrow windows of opportunity make it vitally important to
assess demand as accurately as possible. The most damaging
situation is having a stock shortage while the product is still hot,
leading disappointed consumers to purchase a competitor’s
product.

These sectors make use of sophisticated modelling techniques

developed by academics that use substitution and diffusion rates
to forecast how rapidly new technologies replace older ones. Such
methodologies might not be appropriate to many businesses, but
the message is the same; combining different forecasting
techniques gives more accurate results.

Step 8: Reality check the forecast

Whenever reliable data exists, always check the forecast against
the sales evolution of comparable products to see if it is realistic.
Similarly you should also estimate how your market share might
evolve as new competitors came into this emerging category and
how the total market might grow. Unless this macro overview is
credible, be prepared to rework the assumptions behind the
model.

Step 9: Reforecast, reforecast and reforecast some more

Diligently monitor sales and qualitative feedback such as product
reviews, media mentions, and customer feedback, and agree with
the members of the working group how the assumptions in the
model might need to change. If it’s appropriate, reforecast daily.

Step 10: Be prepared to cut your losses

Finally, always have a contingency plan. A high proportion of new
products fail and it is better to pull the plug on an ailing new
product that is unlikely to achieve a viable level of profitability at
the earliest opportunity. So quantify and agree what level of sales
penetration constitutes failure well before the product launch.
That way, the decision will be swift and the existing stock can be
quickly and cost-efficiently depleted.

Forecasting demand for new products is not an exact science and

relies on judgement rather than statistical techniques. Key to
success are collaboration, using all the quantitative and
qualitative data that is available and having a modelling solution
that can quickly and easily be updated to generate detailed
forecasts for all users across the business. The benefits can be
impressive both in terms of reduced inventory costs and
improved customer satisfaction, something that is vital for a new
product to flourish.