Survey methods of demand forecasting

Survey evidence can improve a demand forecast when respondents possess relevant knowledge and the questions reflect the decision. This article covers customer intentions, salesforce estimates, Delphi panels and pilot test marketing.

When to use it

• Use one or more methods when informed external or frontline judgement would strengthen the The HOOF approach to demand forecasting.

Origins

The methods come from different research traditions. Customer-intention and salesforce surveys developed through market research and sales planning. The Delphi method was created at RAND as a structured way to elicit and refine independent expert judgement. Pilot test marketing grew from product and advertising research. Together they provide complementary evidence rather than one unified forecasting model.

What it is

David Letterman joked that three out of every four people make up 75 per cent of the population. The joke captures the central warning: a numerical survey result can be perfectly calculated and still offer little insight.

The four methods are:

Survey of customers’ intentions

Salesforce estimation

Delphi

Pilot test marketing

How to use it

Survey of customers’ intentions

Select a representative group from each major product/market segment and ask about intended total purchases from all suppliers over the next 12 and 24 months. The questions can accompany a broader interview about purchasing criteria and comparative supplier performance, as shown in Figure B.1.

Purchasing contacts may not know their organisation’s future sales or production plans and may refer the interviewer to sales colleagues. The firm may have a weaker relationship with those people, limiting the evidence. A separate survey of customer sales teams can therefore cost more than it contributes.

Set modest expectations and, where possible, add the intention questions to strategically necessary customer interviews. Treat stated intention as one input: budget authority, uncertainty, optimism and the gap between what people say and do all affect accuracy.

Figure B.1 Survey of customers’ intentions

Bring together knowledgeable salespeople in a workshop or structured online discussion. Ask them independently where market demand is heading, then facilitate a debate and consolidate the reasons behind their estimates.

Salespeople are close to customers and may anticipate purchasing changes. They can also mistake their accounts for the whole market, overlooking customers served by competitors or segments outside their territory. Separate account-level pipeline evidence from market-level judgement, look for shared assumptions and adjust for incentives that encourage optimism or conservatism.

The method is inexpensive and worth testing when the organisation understands the strengths and limits of its salesforce.

The Delphi method

Select a small group of informed industry observers and ask each person the same carefully designed questions independently. Aggregate the responses anonymously and return a summary to the panel. Participants review the group view and either retain or revise their answer, explaining the reason.

Update the summary after each round. Stop when the forecast and reasoning have stabilised enough for the decision, or when another iteration is unlikely to justify the effort. A final draft can be circulated for correction.

Anonymity reduces pressure to defer to a dominant individual, but it does not make the panel representative or correct. Document panel selection, disagreement and the assumptions behind convergence.

Pilot test marketing

Use a pilot when introducing a new product, service or product/market combination and management needs behavioural evidence of demand and price acceptance.

For a business-to-business offer, arrange discussions or limited trials with prospective buyers. Explain the customer benefit and price, record reactions consistently and analyse the evidence. Produce a concise research report in which each conclusion is supported by attributed comments where permission exists, credible third-party evidence or relevant market data.

For a consumer offer, test with people who resemble the target population in a suitable retail or digital environment. Show a product or explain a service briefly and consistently. Record exposure, interest, objections and actual behaviour where possible. Do not present a convenient street sample as statistically representative.

Use responses to estimate potential market size. Imagine broad awareness and competing suppliers, then ask what total demand could exist. Compare the result with adjacent products or services and challenge whether penetration, repeat purchase and scaling assumptions are plausible.

Top practical tip

Combine methods with different biases—for example, customer intention, salesforce knowledge and a small behavioural pilot—and investigate why their estimates disagree.

Top pitfall

Do not convert opinion into apparent certainty. Every method depends on who was asked, what they know, incentives, wording and whether stated intention becomes behaviour.

Further reading

• Rowe, G. and Wright, G. (nineteen ninety-nine). “The Delphi Technique as a Forecasting Tool: Issues and Analysis.” International Journal of Forecasting.
• Armstrong, J.S. (ed.) (two thousand and one). Principles of Forecasting: A Handbook for Researchers and Practitioners. Kluwer Academic Publishers.