What is Demand Forecasting? Demand Forecasting of New Products

Demand Forecasting

Good strategic planning rests on the foundation of good forecasting. Unlike backcasting (backward projection of data), it is a forward projection of data. It is an essential tool in developing new products, scheduling production, determining necessary inventory levels, and creating a distribution system.

Its essence is estimating future events according to past patterns and applying judgment to those projections. In other words, forecasting is an attempt to foresee the future by examining the past. Business firms can estimate and minimize future risk and uncertainty through forecasting and forward planning. Without forecasting, forward planning will be directionless and meaningless.

These ‘ex-ente forecasts’ are often made for a number of periods in the future. The forecasts for past and present periods, which are carried out to test the credibility of the forecasting models, are called ‘ex-post’ forecasts.

As production takes time (called the gestation period), business firms would like to know the likely demand for a product at a future date, to plan and execute its production properly. It assumes greater relevance where demand conditions are more uncertain than supply conditions.

Demand forecasting is an attempt to estimate the future level of demand on the basis of past as well as present knowledge and experience, to avoid both underproduction and overproduction. It may be based on estimates of the demand potential of the entire industry. Demand forecasting serves as the reference point for all marketing control efforts. It is indispensable in modern business.

Demand forecasts may be passive or active. The former predicts future demand by extrapolating the demands of the previous years in the absence of any action by the firm. Here, things are assumed to continue the way they have been in the past. These forecasts are used only to assess the impact of new policies on the market.

The latter estimates the future scenario, inclusive of its own future actions and strategies of the firm itself. These forecasts are more meaningful, as they take into account the likely changes in the relevant variable in estimating future demand. Here, the firm manipulates the demand by changing prices, product quality, promotion efforts, etc.

Demand forecasting is very popular in developed countries, where demand conditions are relatively more uncertain. In India also, economic liberalization and consequent rising competition during recent years have increased the importance of demand forecasting.

That is why the National Council of Applied Economic Research (NCAER) prepares macro demand forecasts for a number of products. These help in forecasting industry demand, company demand, and market segment demand.

Characteristics of a Good Demand Forecasting Method

Eight major characteristics can be identified with forecasting methods (techniques) to identify key characteristics of a good demand forecasting method. The marketing manager is expected to study these characteristics while deciding on a method of demand forecasting:

Time Horizon

The length of time over which a decision is being made has a bearing on the appropriate technique to use. The probability of forecasting error generally decreases with an increase in the length of the time horizon. Long periods, such as more than a year, are difficult to model in a way that these are based on extrapolation of history.

Demand forecasts for consumer durables and plant replacements are generally on long-term time horizons. Whereas, demand forecasts for consumer goods are bound to be of shorter time horizon. Generally speaking, qualitative forecasting procedures are better suited for forecasting long periods in the future. On the other hand, quantitative techniques are more accurate in the short run, such as within a year or less.

Level of Detail

The level of detail needed should match the focus of the decision-making unit in the forecast. For example, production planning must make its decision at the individual product level, whereas the corporate planning department is likely to be happy with aggregate demand forecasts by product categories.

The more items that forecasts are required for, the greater the need to use straightforward models that require little time to implement.


Forecasting in situations that are relatively stable over time requires less attention than those that are in constant flux. In stable situations, the existing pattern is assumed to continue in the future and past patterns can be easily extrapolated in the future.

On the other hand, unstable and uncertain situations require more attention by the management and a greater total forecasting effort, particularly, for the latest information.

Pattern of Data

Data required to use the underlying relationships should be available on a timely basis. Each forecasting method is based on an underlying assumption about the data. As different forecasting methods vary in their ability to identify different patterns, it is useful to make the pattern in the data fit with the method that suits it the most.

To judge whether the model fits the data being forecast, it is imperative to know the assumptions behind the model selected.

Much data is available within the firm. Some appear in records and reports such as annual statements, shipping documents, invoices, employment records, and production reports. The organization might have a Management Information System (MIS) responsible for gathering and preprocessing relevant data.

Other information is often collected from sources outside the firm, including publications by Government (like Central Statistical Organisation), universities, foundations, trade associations, and professional research firms. Government sources are particularly valuable since federal agencies acquire a huge quantity and variety of data and publish the results in an easily available and inexpensive form.

Analysts must strive to ensure that data obtained from others are accurate, precise, and relevant; otherwise demand forecasts will be in error. They should be especially careful to determine that adequate diligence and care were exercised by the source. They may have to examine the definitions of variables used by others.

Because these may not agree with their own. Further, analysts should try to uncover evidence of purposeful distortion of data by others in the light of their own special interests. For example, a trade association or special-cause group might overstate some data to support its position on some matter.

Type of Model

Other assumptions are also made in each forecasting technique that must fit the situation under consideration. To illustrate, regression (one type of technique) assumes causality by one or more independent variables. This causal must be realistic (not just mathematically convenient), for the model to be used.

Management need not be experts in the mathematical details of each method, but they should know the assumptions of any model and whether these fit a particular situation. The technique used should be easily comprehended by the management to give quick meaningful results.


Several costs are associated with adopting forecasting procedures within an organization, like managerial development, storage, operation, and opportunity (in terms of other techniques that might have been applied).

Naturally, the variation in costs affects the selection of the forecasting method. There is a need for an economic consideration of balancing the benefits against the extra cost of providing improved forecasting.


It is measured by the degree of deviations between past forecasts and current actual performance or present forecasts and future performance. If the likely state comes close to the actual state, it means that the forecast is dependable. The level of accuracy required has a bearing on the model to be selected.

A tolerable error might be 10 percent in some instances, whereas a five percent error spells disaster in others. The greater the required accuracy, the greater the cost of generating a forecast. The forecaster has to make a trade-off between the accuracy required and the cost to achieve that accuracy through cost-revenue analysis.

Ease of Application

Models must be chosen within the abilities of the users to understand them and within the time allowed for using them. This will enable management to properly interpret the results. The simplicity of handling the method matters in the selection of the method.

Steps in Demand Forecasting

Demand forecasting is a scientific exercise. It has to go through a number of steps. At each step, critical considerations are required to be made. The following steps are necessary for demand forecasting. These steps present a systematic way of initiating, designing, and implementing a forecasting system.

Identification of Objective

To begin with, the economist should be clear about the uses of forecast data and how it is related to forward planning by the firm. Depending upon its use, the economist has to choose the type of forecast: short-run, active or passive, conditional or non-conditional, etc.

Nature of Product and Market

The important consideration is the nature of the product or service for which we are attempting a demand forecast. We have to examine carefully whether the product is a consumer good or producer good, perishable or durable. While analyzing the demand for finished goods, the demand for corresponding raw materials and intermediate goods should also be analyzed.

The elasticity of demand for intermediate goods depends on their relative importance in the price of the final product. Further, advertising and price-cutting are more important for final goods than intermediate goods.

The time factor is a crucial determinant in demand forecasting. Perishable commodities, such as fresh vegetables and fruits, can only be sold over a limited period of time. If there are storage facilities, consumers can adjust demand according to price, income, and availability.

Here, demand forecasting can avoid waste. Furthermore, the forecasting of demand must consider the stage at which the product is i.e. introduction (slow rise in sales), growth (rapid rise in sales with acceptance of the product), maturity and saturation (maximum sales), or obsolescence and decline (sales taper off with the introduction of substitute products).

Finally, the nature of competition in the market (perfect or imperfect) should not be overlooked.

Determinants of Demand

Depending on the nature of the product and the nature of forecasts, different determinants will assume different degrees of importance in different demand functions. In addition, it is important to consider socio-psychological determinants; especially demographic, sociological, and psychological factors affecting the demand.

Analysis of Factors

In an analysis of the statistical demand function, it is customary to classify the explanatory factors into

  • trend factors,
  • cyclical factors,
  • seasonal factors and
  • random factors.

An analysis of factors is especially important depending on whether it is the aggregate demand in the economy or the industry’s demand or the company’s demand or the consumer’s demand which is being predicted.

Choice of Method

This is a very important step. The economist has to choose a particular technique from among various techniques of demand forecasting, depending on the nature of the product. Then data is collected to make the forecast. In some cases, it may be possible to use more than one method.

However, the choice of method should be logical and appropriate, as accuracy, to a great extent, depends on this choice. The choice itself depends on a number of factors – the degree of accuracy required, reference period of the forecast, complexity of the relationship postulated in the demand function, available time for forecasting exercise, availability of data, size of cost budget for the forecast, etc.

Testing Accuracy

There are various methods for testing statistical accuracy in a given forecast. Some of them are simple and inexpensive; others are quite complex and difficult. This testing is needed to avoid/reduce the margin of forecasting error and thereby improve decision-making.

The ‘absolute level of forecasting error’ is equal to the difference between the actual value and the forecast value. Graphically, it is measured by the vertical distance between the forecast value curve and the 45-degree line (showing perfect accuracy due to the coincidence of forecast value and realized value) for a particular period.

If forecasts are made for more than one year, then the average absolute level of error is found by taking the arithmetic mean of the absolute values of forecasting errors of different periods. However, the Percentage Absolute Error (PAE) test is better, which is mathematically shown as follows:

Demand Forecasting of New Products

Projecting demand for new products is different from those for established products. This requires an intensive study of the economic and competitive characteristics of the product. It also requires probing the mind and resources of the customers through surveys. Forecasting methods need to be tailored to the particular product.

Product Life Cycle Analysis

Many products generally have a characteristic known as ‘perishable distinctiveness’. This means that a product is distinct when it degenerates over the years into a common product. The innovation of a new product and its degeneration into a common product is termed as the life cycle of a product.

The forecaster should identify the phase of the product cycle of a product as shown by the S-shaped curve in Figure. Knowledge about the product life cycle and turning points on the curve, right at the beginning itself, is useful for demand forecasting of a product.


Research or engineering skill leads to product development and the product is formally released in the market. Awareness and acceptance are minimal. There are high promotional costs and low commercial exploitation. Sometimes, a product may generate a new demand. The volume of sales is low and there may be heavy losses. The company has to be very careful, as the chances of dying out are very high.


The product begins to make rapid sales gains because of the cumulative effects of introductory promotion, distribution, and word-of-mouth influence. High and sharply rising profits may be witnessed. But, to sustain growth, consumer satisfaction must be ensured at this stage. Here, the risk of product obsolescence and hence shortening of the product life cycle is great. Product competition is more important till this stage.


Price competition begins after the product is established and reaches the stage of maturity. Sales growth continues, but, at a falling rate because of the declining number of potential customers, who remain unaware of the product or who have taken no action. Further, the last of the unsuccessful competing brands will probably withdraw from the market.

For this reason, sales are likely to continue to rise, while the customers for the withdrawn brands are mopped up by the survivors. There is no improvement in the product, but changes in selling efforts are common. Profit margins slip despite rising sales.


Sales reach and remain on a plateau marked by the level of replacement demand. There is little additional demand to be stimulated. This is shown by the stable portion in Figure.


Sales begin to diminish absolutely, as the customers begin to tire of the product and better products or substitutes gradually edge out the product. For example, mobile phones and diesel cars.

There are several reasons why the life cycle of a product tends to be short:

  • continuous research for product development,
  • simultaneous attempts by several companies in the same direction, and
  • the tendency of a new idea to attract competitors.

Improvements offered by one company are likely to be met and, if possible, exceeded by competitors in a relatively short period. If a competitor hits upon a real improvement (perhaps based on an entirely new technology) and markets it well, both sales and profits of the original product innovator may decline drastically.

It may be noted that products may begin a new cycle or revert to an early stage as a result of

  • the discovery of new uses,
  • the appearance of new users, and
  • introduction of new features.

Test Marketing

Even where the product is of high quality, market failure is still a possibility, if other important factors in the marketing mix show weakness. It is, therefore, logical to examine how the company’s total marketing mix may be tested by conducting test marketing.

Under test marketing, the product is introduced in selected areas, often at different prices. The number of areas selected depends on the representativeness and the cost of marketing. The selected areas must have average competition, the presence of chain and departmental stores, the existence of various types of basic industries, and the optimum size of the population.

The duration of testing depends upon the average purchase period, the competitive situation,, and the cost of testing. It is necessary to collect the necessary information regarding the nature of the product, the nature of customers, channels of distribution, buyers’ behavior, etc.

These tests would provide the management with an idea of the amount and elasticity of the demand for the product, the competition it is likely to face, and the expected sales volume and profits it might yield at different prices. Experience shows that the chances of a new product being successful are ‘significantly greater’ if it is first put into a controlled test market, where it is exposed to realistic competitive conditions.

To make test marketing more fruitful, a ‘post-launching’ survey should be conducted. The survey will reveal whether the earlier satisfaction continues to be derived, whether people like the product and make repurchase, whether the advertising is appealing, etc. On the basis of the findings, changes will have to be incorporated before the product is finally launched in the market.

Survey of Consumers’ Intentions

This method involves interviewing consumers by sending questionnaires to elicit replies as to make short-term predictions of demand. Samples may be given for this purpose. This method (also known as opinion surveys) is most useful when the bulk of the sales is made to industrial producers.

Here, the burden of forecasting is shifted to the customers. Yet, it would not be wise to depend wholly on the consumers’ estimates and they should be used cautiously in the light of the judgment of the specialized dealers or sellers.

Evolutionary Approach

The demand for a new product may be projected as an outgrowth and evolution of an existing old product. This approach is successful when the new product is merely an improvement of an existing product.

Growth Curve Approach

The role of growth and demand for new products may be estimated on the basis of the pattern growth of some established substitute products.

  • Tapan K Panda, Marketing Management, Excel Books.

  • V S Ramaswami and S Namakumari, Marketing Management, Macmillan, 2003.

  • Ramphal and Gupta, Case and Simulations in Marketing, Galgotia

  • Jayachandran, Marketing Management, Tata McGraw Hill, 2003.

Leave a Reply