Testing prices: which method shows what customers will really pay?

A price increase for existing customers, the entry price of a new plan, the premium for a higher-end variant: each of these decisions needs a different kind of evidence. Choosing the method to fit the question saves time and avoids false precision.

Van Westendorp: the accepted price range

The price sensitivity meter asks for four price thresholds: when a product seems too cheap, cheap, expensive and too expensive. The answers produce a range of acceptable prices. The method is fast and inexpensive and well suited to early phases. But it shows no demand per price, and respondents see neither alternatives nor a real purchase situation.

Gabor-Granger: willingness to buy per price point

Respondents see different prices one after another and state for each whether they would buy. From this, a price-demand curve can be estimated. The drawback: anyone who sees several prices in a row answers strategically, and willingness to buy remains a statement without consequence.

Conjoint: price in combination with many attributes

Conjoint analyses have respondents choose between product profiles that differ in several attributes, including price. This makes it possible to estimate how much individual attributes are worth. Conjoint is strong when many attributes need to be weighted at once. The choice remains hypothetical, and the results depend heavily on the design and the selection of attributes.

MaxDiff: what matters, not what it may cost

MaxDiff is not a pricing method, but it is often mentioned in the same breath. Respondents choose the most and least important attribute from lists. This produces a clean ranking of benefits and arguments, for example for packaging. MaxDiff does not provide a willingness to pay.

Price test with real purchase behaviour

Comparable groups from your target audience see the same offer in a realistic ad and on a realistic offer page, each group at a different price. What is measured is the purchase or sign-up decision. Nothing is sold and no payment is taken; anyone who decides learns straight away that the offer is not available yet. The result is a price-demand curve based on behaviour rather than statements.

• Van Westendorp: Answers: Which price range is considered acceptable? · Data basis: Stated
• Gabor-Granger: Answers: How does willingness to buy change per price? · Data basis: Stated
• Conjoint: Answers: How much are individual attributes worth? · Data basis: Stated, as a choice
• MaxDiff: Answers: Which attributes matter most? · Data basis: Stated, as a ranking
• Painted door price test: Answers: At which price do customers buy, and which price earns the most? · Data basis: Observed

Which method when?

• Early phase, no price idea yet: Van Westendorp for the range.
• Many attributes and packages to weight: conjoint, complemented by MaxDiff for the arguments.
• Two to six specific prices, and budget depends on the decision: a price test with real purchase behaviour.
• Price increase for existing offers: a price test with the current and the new price, so you see how much demand the increase costs.

Cost-based calculation remains the lower limit: it shows which price pays off, but not which price customers accept. That is why we ask for your minimum margin before every price test, so no variant is tested that you could not implement anyway.

Limits of the behavioural test

A price test compares a few specific prices and shows what people do, not why. How competitors react, or how existing customers respond to an increase, you look at separately. Surveys and behavioural measurement are therefore not opposites: one explains, the other decides.