How to Measure Willingness to Pay: 5 Methods for Founders

You can measure your customers' willingness to pay in five practical ways: structured customer interviews, the Van Westendorp method, Gabor-Granger price tests, real sales conversations, and an analysis of the budget your target customer spends on alternatives today. None of these methods gives you an exact number. Together they give you something more important: an evidence-backed pricing corridor instead of a guess.

The uncomfortable truth first: what people say in a survey and what they later pay are not the same thing. That is why every method below has a "Limitations" section. Ignore those limitations and you have only replaced gut feeling with gut feeling plus a chart.

What willingness to pay actually is

Willingness to pay is the highest price a specific customer is prepared to pay for your product. The key words are "a specific customer": it is not a property of your product but of the person across the table. The same tool is worth completely different amounts to a 5-person startup and to a 500-person mid-sized company, and exactly that spread later becomes the basis for your price tiers.

Method 1: Structured customer interviews

The most direct route, and the only one that also tells you why. The art lies in asking about value, not about price.

How to do it: Talk to 8 to 12 people from your target group. Do not ask "What would you pay?" Reconstruct the status quo instead:

1. "How do you solve this problem today?"
2. "How much time does that cost you per week, at an honest estimate?"
3. "What was the last tool or service you bought for this area, and what did it cost?"
4. "If this problem were gone tomorrow: what would actually change in your week?"
5. "Who in your company would have to sign off on a purchase like this, and at what amount does it get complicated?"

Questions 3 and 5 are the most valuable. Question 3 gives you real amounts that have already been spent, rather than hypothetical ones. Question 5 uncovers the approval threshold where many B2B deals fail, regardless of the perceived benefit.

Limitations: People underestimate how much time they spend on a problem and overestimate their willingness to buy, especially when talking to likeable founders. Treat statements like "I would buy that straight away" as interest, not revenue.

Method 2: The Van Westendorp method

The Price Sensitivity Meter by Peter van Westendorp, presented at the ESOMAR congress in 1976, asks four questions about price perception instead of one question about price:

1. At what price would the product be so cheap that you would doubt its quality?
2. Up to what price would it be a good deal?
3. At what price does it become expensive, but still justifiable?
4. At what price would it be so expensive that buying it is out of the question?

The answers from all respondents produce four curves. Their intersections give you a range of acceptable prices, the optimal price point (OPP) and the indifference price point (IPP). The full explanation is in the glossary under Van Westendorp method. With the free Van Westendorp tool you can set up the survey in a minute and get the chart automatically.

Limitations, and this is where it gets uncomfortable: For robust results, market research practice recommends samples of roughly 200 to 400 respondents in consumer markets and around 50 to 100 in B2B, with about 100 completed responses as the lower bound for reliable conclusions (SIS International, Conjointly). As a pre-revenue founder, you will not reach these numbers.

That is no reason to throw the method out, but it is a reason to read it honestly. With 15 to 30 responses from your real target group you get a direction, not precision. In that case, treat the OPP as the midpoint of a rough corridor, not as a price recommendation to two decimal places. And one methodological trap: if your respondents do not have a clear picture of the product, you are not measuring their willingness to pay, you are measuring their imagination. Always describe the product and the price basis (per month, one-off, per user) above the four questions.

Method 3: Gabor-Granger price test

Here the respondent is shown a specific price and asked whether they would buy at that price. If they say yes, the price goes up; if they say no, it comes down, until the willingness to buy flips. Across all respondents this produces a demand curve, and with it an estimate of the revenue-maximising price.

When it makes sense: When you already have a price range and want to sharpen it. Gabor-Granger answers "which of these prices", not "what ballpark are we even in".

Limitations: The method measures willingness to buy without any competitive context. Ask someone about a price in isolation and they block out the alternatives that would be front of mind straight away in a real buying process.

Method 4: The real sales conversation

The only method that deals in hard currency. Name a specific price in the conversation and watch what happens. Not what is said, but what follows: do they ask about the contract, the discount, the sign-off? Or does the topic get politely postponed?

A practical approach for your first customers: deliberately start at the top of your assumed corridor. A no at a high price costs you one prospect. A yes that comes too quickly at a low price costs you your price position for every customer who follows, and winning that back is considerably more expensive.

Limitations: Small sample sizes, and strong bias from the way you run the conversation. Only count commitments once money has changed hands.

Method 5: Budget analysis instead of a survey

The most underrated route. Instead of asking what someone would pay, you research what they already pay today: for competing products, for tools in the same category, for the internal working time your product replaces.

These figures are often public, in price lists, job ads (what does the person whose time you save cost?) or public tenders. They are more robust than any survey because they reflect money that has actually been spent. In PricingOS this exact logic sits in steps 2 and 3: the ideal customer's budget and the real prices of the alternatives.

Limitations: You are measuring the status quo, not the willingness to pay for something significantly better. If you are redefining a category, existing budgets only show you the lower bound.

Which method when?

From signal to price

None of these methods gives you your price. They give you the upper limit of your corridor. The lower limit comes from your cost calculation, which you can work out in two minutes with the break-even calculator. Between the price floor and the price ceiling lies your pricing corridor, and only inside that corridor does the price become a strategic decision rather than a guess.

How these building blocks fit together into a complete pricing strategy is covered in the guide How to Build a Pricing Strategy: The 9-Step Process for Founders. Willingness to pay is part of step 6 there.

Common questions

How many responses do I need at minimum? For a direction, 15 to 30 responses from your real target group are enough. For robust Van Westendorp results, market research cites much higher numbers (around 50 to 100 in B2B, 200 to 400 in consumer markets). Be clear internally about which of the two you actually have in hand.

Can I survey my existing customers? Yes, and they are in fact the best sample you can get: they know the product and have already paid. Just bear in mind that they know your current price, which pulls their answers towards it.

What about A/B tests on the pricing page? Most early B2B products simply lack the traffic for meaningful results, and showing different prices for the same service at the same time also raises questions of fairness towards customers. At this stage, interviews and sales conversations deliver more per hour invested.

What if the answers are far apart? Then you do not have bad data, you have a segmentation signal. A wide spread usually means you are looking at several customer groups who get different value from your product, and that is exactly where price tiers come from later.