Demand forecasting: meaning, types, steps and survey methods (including Delphi)

1. Meaning

Demand forecasting is the estimation of the future demand for a firm’s product over a given period, on the basis of past data and present conditions. It is a scientific estimate, not a guess: it rests on the factors that determine demand and on a stated method. Every plan of a firm — production, purchase of materials, stock, manpower, finance and expansion — starts from the forecast of sales.

2. Objectives

Short-term objectives (up to a year): production scheduling so that output matches sales; buying raw materials in time; keeping stock at the right level; fixing a pricing policy for the season; setting sales targets and controlling the sales force; arranging working capital.

Long-term objectives (beyond a year): planning a new unit or expansion of plant; long-term finance; manpower planning and training; product development and entry into new markets.

3. Types (levels) of demand forecasting

• By period: short-term forecasting (seasonal and current) and long-term forecasting (trend, for capacity planning).
• By level: macro forecasting (demand in the whole economy, from national income and aggregate spending); industry forecasting (demand for the output of the whole industry, often by trade associations); firm-level forecasting (demand for one firm’s product, the most useful to management).
• By nature of the forecast: passive forecasting, which assumes the firm does not change its own policy, and active forecasting, which allows for planned changes in price, advertising or product.
• By product: forecasting for an established product (past data exist) and for a new product (no past data; requires survey, test marketing or analogy with similar products).
• By type of good: consumer goods, durable goods (where replacement demand must be estimated separately) and capital goods (derived demand).

4. Steps in demand forecasting

1. Fix the objective: what is to be forecast, for what period and at what level.
2. Identify the determinants of demand for the product.
3. Choose the method of forecasting.
4. Collect the data (primary or secondary).
5. Analyse the data and make the estimate.
6. Interpret the result and check it against actual sales as they come in.

5. Methods: an overview

6. Survey methods

(a) Consumer survey method. Buyers are asked directly what they intend to buy.

• Complete enumeration (census): every potential buyer is asked. Accurate but costly and feasible only where buyers are few.
• Sample survey: a representative sample is asked and the result is blown up to the whole market. Cheaper and quicker; accuracy depends on the sample.
• End-use method: for intermediate goods, the forecast is built up from the planned output of the industries that use the product (demand for steel from planned output of construction, automobiles and machinery).

(b) Sales force opinion (collective opinion) method. Salesmen estimate likely sales in their territories; the estimates are combined and adjusted by the sales manager. Quick and based on first-hand knowledge, but salesmen may understate to keep targets low, or be over-optimistic.

(c) Expert opinion method. Specialists inside or outside the firm (dealers, consultants, trade bodies) give their estimates, which are pooled.

(d) Delphi method. A refined form of expert opinion, developed at the RAND Corporation in the United States in the 1950s. Its steps:

1. A panel of experts is chosen; they do not meet and are kept anonymous to one another.
2. Each gives a forecast through a questionnaire, with reasons.
3. A coordinator summarises the replies (range, median, reasons) and sends the summary back.
4. Experts revise their estimates in the light of the others’ views.
5. The rounds are repeated until the estimates converge on a consensus.

Merits: anonymity removes the pressure of a dominant member and of group thinking; experts in different places can take part; useful for new products and long-term forecasts where data are lacking. Demerits: slow; depends on the choice of experts and the skill of the coordinator; experts may drop out; the consensus may be forced.

(e) Market experiment (test marketing). The product is sold in a few representative markets and buyers’ response to different prices, packages or advertisements is observed. Real behaviour rather than stated intention, but costly and slow, and competitors may disturb the test.

Merits of survey methods generally: first-hand information; useful where no past data exist (new products); reflects current intentions. Demerits: buyers’ stated intentions often differ from what they actually do; costly; subjective; less suited to long-term forecasts.

7. Characteristics (criteria) of a good forecast

1. Accuracy: the forecast should be close to actual sales.
2. Simplicity: management should understand the method.
3. Economy: the cost should be justified by the benefit.
4. Availability of data: the method should use data that can be obtained.
5. Timeliness: it should be ready when the decision has to be made.
6. Flexibility: it should adapt to changing conditions.
7. Durability: the relationships it relies on should hold for a reasonable time.
8. Plausibility: the result should make sense to management.

How this is examined

January 2026: “What is demand forecasting? Explain its types” (10 marks). August 2024: “What is demand forecasting? Explain the survey method” (10 marks) and “Characteristics of good demand forecasting” (6 marks). February 2025: “Delphi method” (6 marks). A 10-mark answer should define forecasting, give its objectives, its types or the methods table, then the method asked for in detail with merits and demerits.