How to Measure Willingness to Pay Before You Ship

Three rules keep the tests honest.

Recruit real buyers, not colleagues or fans. People who already like you, or who will never open their wallet, distort every metric. Recruit from the same segment you plan to sell to, and make sure they have the problem your product solves. Salesforce's guide to value-based pricing recommends segmenting your market by industry, company size, location or need, then interviewing representatives from each group. Do the same for test recruitment.

Never ask the direct question. "How much would you pay?" is close to useless. People lowball it to nudge you toward a lower price, or they answer from emotion with no real anchor. The Price Sensitivity Meter was built specifically to avoid that failure by asking about price perceptions around a product rather than a single figure, per the method's overview on Wikipedia. A presentation from POCTRN is blunter: direct questioning is "generally meaningless," and laddering questions ("Would you buy at $10? What about $8?") should not be used at all (POCTRN PDF).

Decide the pass threshold before you run. Write down the click rate, the pre-sale conversion or the revenue figure that would justify building. Koji's fake door guide makes this an explicit step: a click rate only means something relative to a threshold you set in advance. Deciding after you see the data is how teams talk themselves into shipping.

The Van Westendorp Price Sensitivity Meter, introduced by Dutch economist Peter van Westendorp in 1976, asks four questions about price perception. Each respondent answers all four, and you plot the results as four cumulative curves, per Wikipedia. Here is the standard wording, adapted from SurveyMonkey's walkthrough and Quantilope's question examples:

1. At what price would the product be so expensive that you would not consider buying it? (Too expensive)
2. At what price would it be so cheap that you would question the quality? (Too cheap)
3. At what price does it start to get expensive, so you would have to think about it, but it is not out of the question? (Expensive)
4. At what price would it be a bargain, a great buy for the money? (Cheap)

Design the survey

Show respondents a clear description of the product before the questions. Quantilope recommends setting the stage with a product outline so people can visualize what they are pricing. Use open numeric fields rather than preset price buckets; forcing respondents to pick from a list loses precision.

Recruit enough respondents that the curves are stable. The method was originally used with small samples in interviews and focus groups, per POCTRN, but larger samples give smoother distributions. If your buyers split into very different segments, run the analysis separately for each one.

Read the four intersections

Plot cumulative frequencies for each question. Invert the "too cheap" and "cheap" curves so all four can cross, then read these points, per Conjointly's tool documentation:

For a worked example, Conjointly shows a chart where the acceptable range runs from $5 to $10 with an "optimal" price of $6.75 (Conjointly). Treat that optimal point as a center of gravity, not a recommendation. The whole chart is the output; a single number pulled from it is a fiction.

Add a purchase-intent extension

The basic meter tells you perception, not demand. The Newton, Miller and Smith extension adds two follow-up questions about how likely the respondent is to buy at the expensive and cheap prices, which lets you model a rough revenue curve, per Wikipedia and Sawtooth Software. Without this step the survey cannot say anything about volume, and volume is what decides whether a price is profitable. A high Price Sensitivity count is not the same as a profitable price; POCTRN's deck makes the point that higher volume is not necessarily higher profit.

Know the limits

The method assumes respondents can imagine a pricing landscape and that price is an intrinsic signal of value (Wikipedia). It works best when the product is new enough that no strong reference price exists, per Sawtooth. For products in a mature category with clear comparables, monadic pricing questions or conjoint analysis usually give better answers, Sawtooth and Conjointly both note. And remember the hypothetical bias: the numbers come out high compared with real choices.

A fake door test (also called a painted door test) shows users a real-looking entry point, such as a button, a menu item or a pricing tier, that does not lead to a working feature yet. You measure who clicks and how many try to use it, per Koji's guide. The reason to run one is straightforward: a large share of software features go unused. The Standish Group's CHAOS research put rarely or never used features at 64%, and a later Pendo analysis put the figure at 80%, both cited in Koji's guide.

Design the test

• Build a convincing entry point. For a pricing test, that means a real buy button with a real price on it, not a vague "Learn more."
• Instrument the click. Track who clicks, from which segment, and how often. Koji lists click-through rate as the headline metric, with segment, repeat intent and opt-in rate as the supporting signals.
• Be honest at the door. Show a clear "coming soon" message the moment someone clicks, and offer a real way to be notified. Userpilot is direct about the risk: clicks measure curiosity, not commitment, and a badly handled reveal damages trust.
• Do not stack tests on the same people. Userpilot recommends spacing tests at least six months apart per user segment so people do not keep hitting dead ends.

What data to collect

Capture four numbers per price point or segment: the share of exposed users who clicked, the repeat click rate, the opt-in rate (how many left an email for early access), and the distribution of clicks by segment. A price that draws clicks from one segment and none from another changes your business case, so split the data.

Read it conservatively

A click is not a purchase. Userpilot warns that a click does not prove someone would change their workflow, pay more or use the product consistently. Apply a discount factor to your click rate when you build a forecast, and treat the opt-in rate as the more meaningful number because it costs the user a small commitment. Koji's guide also flags the method's biggest weakness: a door tells you how many people want something, not why, so follow up with the clickers (Koji).

The pre-sale is the closest you can get to the real market without shipping. Someone enters card details, gets charged, or gets a formal reservation that commits real money. Everything before this step is an estimate; this step is behavior.

Design it the same way you would design the real purchase. Show the actual price, the actual terms, the actual delivery date you can meet. If you are discounting for early commitment, state that plainly and set a deadline. Then collect three things: how many qualified visitors placed an order, how many of those paid versus abandoned at the payment step, and how much revenue the orders represent.

Interpret it as a floor, not a ceiling. A pre-sale confirms demand at your price, but it often comes from your warmest audience, so it will not tell you how a cold market behaves. It also tells you nothing about long-term retention, which for subscription products matters more than the first payment. A customer's lifetime value shapes what you can afford to spend to acquire them, so model a conservative retention assumption alongside the pre-sale data (Pragmatic Institute).

If the pre-sale fails and the survey passed, the pre-sale wins. Do not rationalize a weak pre-sale by pointing to a good survey; that is the exact gap the hypothetical bias predicts.

Read them together as a ladder. The survey tells you where to point. The door tells you whether the price catches anyone's eye. The pre-sale tells you whether anyone will pay. A result that survives all three is as close to evidence as you will get before you ship. Koji's guide frames it well: build the cheapest experiment that generates real evidence, then decide.

• Leading with the price. Showing a price before describing the product drags every answer toward cheap. Describe the value first, then ask.
• Quoting the optimal price as a recommendation. The OPP is one intersection on a chart, not a mandate (Conjointly).
• Counting clicks as demand. Without a discount, a fake door will overstate what people will pay.
• Citing a competitor's price as your answer. Competitor and cost-plus pricing both ignore what buyers actually value, Maxio's pricing guide lays out why value-based pricing, despite being more research-intensive, is worth the effort.
• Pricing on one metric and calling it done. A single value metric decouples price from value for customers who get very different value per unit, Robin Landy's analysis explains the risk and points to using a second metric or add-ons as a lever.