Pricing research surveys: Van Westendorp, Gabor-Granger, conjoint, and MaxDiff
A practical guide to the four workhorse pricing research methods: how each one works, when to use which, the sample sizes and question formats they need, and the mistakes that quietly ruin the results.
Why "What would you pay?" is the one question you should never ask
Ask people directly what they would pay for your product and you will get a polite fiction. Respondents anchor low because they suspect the answer feeds a price list, they have no real trade-off in front of them, and stated willingness to pay routinely lands 20 to 40 percent below observed behavior. Every serious pricing method exists to get around this problem by asking indirect questions and inferring the answer. The four workhorses are Van Westendorp, Gabor-Granger, conjoint analysis, and MaxDiff. They answer different questions, and picking the wrong one is the most common pricing research mistake.
Van Westendorp: finding the acceptable price range
The Price Sensitivity Meter asks four open-ended price questions about a described product: at what price would it be so cheap you would doubt its quality, at what price would it be a bargain, at what price would it start to feel expensive, and at what price would it be too expensive to consider. Plot the four cumulative curves and their intersections give you a range of acceptable prices, a point of marginal cheapness, a point of marginal expensiveness, and an optimal price point where "too cheap" and "too expensive" cross.
Use it when you have a genuinely new product with no reference price, or you need a defensible range before a more precise study. It is cheap, needs only 100 to 200 respondents per segment, and takes two minutes of survey time.
Common mistakes: running it on a product category where a strong reference price already exists (respondents just echo the market price back at you), describing the product too vaguely for respondents to value it, and reading the "optimal price point" as a recommendation rather than the center of a range. Van Westendorp gives you a range, not a demand curve.
Gabor-Granger: estimating demand at specific price points
Gabor-Granger shows the respondent a product at a specific price and asks purchase likelihood. Depending on the answer, the next question tests a higher or lower price from a predefined ladder. The output is a demand curve: percentage willing to buy at each tested price, from which you can compute a revenue-maximizing price.
Use it when you already know roughly where the price should sit and need to choose between candidate points, for example 49, 59, or 69. It is the natural follow-up to a Van Westendorp study: use the range from one to set the ladder for the other.
Common mistakes: testing price points that are too far apart to interpolate between, starting every respondent at the same price (which anchors the whole sample), and forgetting that the method measures stated demand for a product in isolation, with no competitors on the shelf.
Conjoint analysis: pricing inside a competitive trade-off
Choice-based conjoint is the heavyweight. You decompose the product into attributes and levels, price being one attribute among several, and show respondents repeated choice tasks: three or four product profiles, pick the one you would buy, or none. From hundreds of these choices across the sample, the analysis estimates a utility for every level of every attribute. Because price is traded off against features and brand, you get willingness to pay that is grounded in decisions rather than declarations, and you can simulate market share for any product configuration at any price against specific competitors.
Use it when the pricing question is really a packaging question: which features belong in which tier, at what price, against which competitors. Plan for 200 to 400 respondents minimum, a carefully pruned attribute list (six or seven attributes at most), and real design effort up front.
Common mistakes: overloading the design with attributes until respondents start simplifying and just pick on price, using unrealistic level combinations, and skipping the "none" option so that the model never learns when people would walk away.
MaxDiff: ranking what matters before you price it
MaxDiff, or best-worst scaling, shows sets of four or five items and asks which is most and least important. It produces a ratio-scaled importance ranking that is far more discriminating than rating scales, where everything ends up "very important." MaxDiff does not output a price. Its job in pricing work is upstream: identifying which features actually drive preference, so your conjoint design spends its limited attribute slots on the things that matter, or so you know which features justify a premium tier at all.
Common mistake: treating MaxDiff importance scores as willingness to pay. An item can be universally important and still command no premium because every competitor has it.
Choosing between them
- New product, no reference price, small budget: Van Westendorp for the range, then Gabor-Granger to pick the point.
- Established category, tier and packaging decisions, competitive market: MaxDiff to shortlist features, then choice-based conjoint with a market simulator.
- Quick price check on a defined offer: Gabor-Granger alone.
The methods are complements, not rivals. The most credible pricing studies sequence two of them: a cheap directional method to frame the range, then a trade-off method to make the decision.
Running these without a statistics team
The historical barrier to conjoint and MaxDiff was tooling: experimental design, utility estimation, and simulation used to require specialist software and a consultant. That barrier has mostly fallen. Sinova 360 ships all four methods as native question types: Van Westendorp and Gabor-Granger with dedicated results tabs that draw the intersection charts and demand curves for you, MaxDiff with automated design and scoring, and choice-based conjoint that generates the experimental design, estimates utilities, and includes a market simulator for testing product configurations and price points against competitor profiles. Sinova360 also includes TURF analysis for line optimization and quota controls to keep your sample composition honest. You still own the thinking, especially attribute selection and price ladder construction, but the mechanics no longer require a second tool.
Whichever platform you use, write the study backwards: start from the pricing decision you need to make, pick the lightest method that can make it, and pilot with 20 respondents before fielding. Most pricing study failures are visible in the first 20 completes.