Van Westendorp Price Sensitivity Meter: The 4 Questions

What the Price Sensitivity Meter is

The Van Westendorp Price Sensitivity Meter is a survey technique for determining consumer price preferences. It was introduced in 1976 by the Dutch economist Peter van Westendorp, and has been used since across market research, latterly including software and subscription pricing.

Respondents are asked four questions about the price of a product: the price at which it would be too expensive to consider, expensive but still worth considering, a bargain, and so cheap they would doubt its quality. Each question is plotted as a cumulative curve across the price range, and the points where those curves cross define the boundaries of what the market treats as an acceptable price.

Its distinguishing feature is that it never asks “what would you pay?”, a question that reliably produces useless answers. It instead locates the boundary between cheap enough to be suspect and expensive enough to be out, and reports the corridor between them. The output is a range rather than a single number, which is both the method’s honesty and its main limitation.

Every number and chart below comes from one worked dataset: a calendar scheduling tool for freelancers, 100 respondents screened to people who charge for their time, billed monthly.

Why it matters

Most founders price one of three ways: what the closest competitor charges, what feels right, or what they wish they could charge. None of those is a pricing strategy. They are guesses wearing professional clothing, and the cost of guessing is measurable.

The competitor-matching one deserves a special mention, because it feels like research. It is not. It is inheriting somebody else’s guess, made at a different stage, for a different cost base, possibly by a founder just as uncomfortable as you are.

Price is also the hardest decision to revisit. A published price sets an anchor with every buyer who has already seen it, and raising it later costs goodwill that setting it correctly once would not have.

We can be more precise than “founders price badly”, because we have watched software do it at scale. Asked to price 173 different products, ShipFit’s own pricing stage converged on very nearly the same answer every time. Van Westendorp exists to stop exactly this, and it only works if somebody actually runs it.

When to run it

The four questions

Ask these four, in randomized order, with a one-line product description above them and a single billing frame. Force a single number, not a range.

The wording matters more than founders expect. “So cheap you would question the quality” is doing specific work: it looks for the point where a low price stops signalling value and starts signalling that something is wrong. If your category has no such point, that question returns noise, and you should read the when this is the wrong tool section before you run anything.

What the four curves look like

Each question becomes one cumulative curve. Where they cross is the whole output of the method.

The same figure as numbers, which is what you would build in a spreadsheet:

Cumulative share of 100 respondents. “Too cheap” and “Bargain” fall as price rises; “Expensive” and “Too expensive” climb.

The four intersections

This is the part most write-ups state incorrectly, so it is worth being precise about which two curves produce each point.

Two things follow from that table that founders routinely get backwards.

The acceptable range is PMC to PME, not OPP to IPP. The floor and the ceiling come from the pairs that mix an extreme curve with a moderate one. OPP and IPP are both interior points.

The Optimal Price Point is not the revenue-optimal price. It is the price with the least combined objection. Those are different questions, and the gap between them is covered further down.

How to plot it yourself

Two of the four series accumulate downward and two upward. This is the step that quietly ruins most spreadsheet attempts, because getting one backwards still produces four plausible curves that still cross.

Concretely, for each price point in your grid:

• Too cheap and Bargain: count the respondents whose answer was at or above that price, as a percentage of the sample.
• Expensive and Too expensive: count the respondents whose answer was at or below that price.

Plot all four as lines on one chart, read off the crossings, and you are done. If you would rather not build it, the pricing strategy calculator takes the four median answers and returns the same points.

Clean the data before you plot it

Van Westendorp has exactly one hard validity rule: a respondent’s four answers must be in non-decreasing order. A row that breaks it did not give you a weak signal, it misread a question, and leaving it in pulls every curve toward the middle and narrows your range artificially.

Report the number you dropped alongside your result. A range built from 87 clean responses out of 100 is a more credible claim than one built from “100 responses” that quietly includes 13 incoherent rows.

How many respondents you actually need

The market-research literature says 150 to 300 per segment, and that below roughly 100 the intersections wobble every time you add data. Pre-launch founders almost never have that. Both facts are true at once, so the useful framing is not a single threshold but what a given sample entitles you to claim.

If you cannot get 30 people in your target ICP to answer four questions, that is worth noticing on its own. It is usually a demand signal, not a recruiting problem.

Getting a revenue-maximizing price

The standard criticism of Van Westendorp is that it never considers revenue, and it is correct. The method measures perception, so nothing in its output is weighted by who would actually buy. The Newton, Miller and Smith extension is the accepted fix: add two follow-up questions asking how likely the respondent would be to purchase at their own bargain price and at their own expensive price.

Price multiplied by purchase probability gives you a demand curve and a revenue curve.

This is the single most valuable addition to the classic four questions, and it costs two extra survey fields.

Van Westendorp vs the alternatives

The honest trade is that this method is cheap and its evidence is weak. Everything stronger costs more, needs traffic you may not have, or both.

The sequence matters more than the choice. Van Westendorp narrows an infinite space to a corridor for the cost of a Tally form. Gabor-Granger finds the peak inside that corridor. A live test confirms it with money.

When Van Westendorp is the wrong tool

There is a substantial literature arguing this method should almost never be used. The critiques are not mostly about execution, they are about fit, and four questions settle whether it applies to you.

None of this makes the framework useless. It makes it a cheap first instrument whose output is a hypothesis, which is exactly how the sequence above treats it.

Pricing that isn’t one monthly number

Van Westendorp asks the respondent to price one thing. That was straightforward when SaaS meant a seat and a monthly fee. It is much harder now that a growing share of products bill on consumption or on outcomes, because the buyer has to price a unit they cannot picture.

Two workarounds, in order of preference:

1. Anchor on a representative bundle. Do not ask them to price a credit or an API call. Ask them to price a typical month: “a month where your team schedules around 200 meetings”. They can form a value judgement about that, and you can convert the answer back into unit pricing afterwards.
2. Run one pass per tier. If you already know the shape of your packaging, treat each tier as its own product with its own four questions. You will need the sample size per tier, not across all of them.

For a pure free tier there is nothing to run: no price means no price perception. Run it on the first paid tier.

ShipFit and Van Westendorp

ShipFit applies Van Westendorp at the “How to Charge” stage of the 9-question flow. Feed in your buyer and product description, and the tool produces a starter range with the reasoning written out. If you have real survey data, you input it; if you don’t, you get a first-pass range to validate. Either way, you leave with a price you can defend rather than one you pulled out of a hat.

Where this sits in the sequence

Pricing research fails most often for a reason that has nothing to do with pricing: it was run before anyone confirmed the problem was real, so respondents were pricing a product they had no reason to want.

Further reading

• Peter van Westendorp, NSS. Price Sensitivity Meter (PSM), ESOMAR Congress, 1976 (the original paper)
• The Mom Test. The framework you use before this one, to make sure you’re pricing something people actually need
• Jobs to be Done. Pairs with Van Westendorp to understand what your buyer is paying for versus what your product does
• Pricing strategy calculator. Runs the four price points and returns your defensible range
• CAC / LTV ratio calculator. A price inside the range still fails if acquisition costs more than the customer returns
• Pricing validation. The full stage this framework sits inside
• ARR / MRR. The recurring-revenue numbers your chosen price actually moves