What is the Van Westendorp Price Sensitivity Meter?
Written by Arnon Shimoni
✓ Expert
Last updated on:
The Van Westendorp Price Sensitivity Meter (PSM) is a survey technique that finds an acceptable price range by asking respondents four open-ended questions about price rather than asking them to accept or reject a specific number. Dutch economist Peter van Westendorp introduced it in 1976. The four answers are plotted as cumulative distributions, and the intersections of those curves define a range of credible prices plus two reference points, the Indifference Price Point and the Optimal Price Point.
Field | Detail |
|---|---|
Also known as | PSM, Price Sensitivity Meter, Van Westendorp analysis |
Introduced by | Peter van Westendorp, 1976 |
Question format | 4 open-ended price questions, no price shown to the respondent |
Output | An acceptable price range, plus IPP and OPP reference points |
Does it give a demand curve? | No. Not without the Newton-Miller-Smith extension |
Typical sample | 150-300 respondents per segment (practitioner convention) |
Best for | New products, new categories, cases with no established reference price |
Weakest for | Established categories, usage-based pricing, anything with a variable bill |
Common companion method | Gabor-Granger or conjoint to set the level within the range |
The four questions
The wording matters. Changing it changes the curves. The standard formulation:
Too cheap. "At what price would you consider the product to be priced so low that you would feel the quality couldn't be very good?"
Cheap / a bargain. "At what price would you consider the product to be a bargain, a great buy for the money?"
Expensive. "At what price would you consider the product starting to get expensive, so that it is not out of the question, but you would have to give some thought to buying it?"
Too expensive. "At what price would you consider the product to be so expensive that you would not consider buying it?"
Two things about this design. First, no price is ever shown to the respondent, so there's no anchor to react against, which is what makes PSM usable when nobody knows what the category costs. Second, the "too cheap" question is doing real work: it captures the point where a low price reads as a quality signal in reverse. That's a genuine effect in B2B software, where a €99 enterprise plan makes procurement suspicious rather than happy.
Before you ask any of them, describe the product. The whole study is only as good as that description. Vague concept, vague prices.
How to read the curves
Each of the four answers becomes a cumulative frequency curve. "Too cheap" and "cheap" are plotted as descending cumulatives (the share of respondents who'd say the price is at least that cheap), while "expensive" and "too expensive" are plotted as ascending cumulatives. Four curves, four meaningful crossings.
Intersection | Curves that cross | What it means |
|---|---|---|
Point of Marginal Cheapness (PMC) | Too cheap × Expensive | Lower bound. Below this, quality concerns outweigh bargain appeal |
Point of Marginal Expensiveness (PME) | Too expensive × Cheap | Upper bound. Above this, you're losing more buyers than you're gaining margin from |
Indifference Price Point (IPP) | Cheap × Expensive | Equal numbers call it cheap and call it expensive. Often reads as the market's perceived "normal" price, or as a price leader's position |
Optimal Price Point (OPP) | Too cheap × Too expensive | The point where the two rejection groups are equally sized |
PMC to PME is the range of acceptable pricing. That range is the actual deliverable of a PSM study.
The OPP is the number everyone quotes and the number most often misused. It is not the revenue-maximising price, and it is not the profit-maximising price. It's the price at which the count of people rejecting you for being too expensive equals the count rejecting you for being too cheap. Nothing about that definition involves volume or margin. Calling it "optimal" was, probably, a naming decision van Westendorp would take back if he could.
A useful diagnostic: if PMC and PME sit close together, your respondents agree on what the product is, and you have a tight, confident range. If the range is wide, you're either talking to multiple segments who value the product very differently (split the sample) or your product description was unclear (rerun it).
The Newton-Miller-Smith extension
Plain PSM tells you nothing about how many units you'd sell. The Newton-Miller-Smith extension bolts demand onto it by asking two follow-up purchase-intent questions, using each respondent's own answers as the prices:
"At [the price you called expensive], how likely are you to purchase in the next six months?" (5-point scale)
"At [the price you called a bargain], how likely are you to purchase in the next six months?" (5-point scale)
Combining the intent scores with the price distributions produces a trial-rate curve and, multiplied out, a revenue curve. That gives you the revenue-maximising price the base method can't.
If you're going to run PSM at all, run it with the extension. The marginal cost is two questions. The marginal value is the difference between "here's a range" and "here's a range plus where revenue peaks inside it."
When to use Van Westendorp (and when not to)
Use it when there's no reference price. A genuinely new category, a product with no direct substitute, or a market where the incumbents don't publish prices. PSM is the only common method that doesn't require the respondent to already have a price in their head.
Use it early. It's cheap, it's fast, and it fits a concept description rather than a working product. It's a good first pass before you spend real money on conjoint.
Use it to bound a debate. When the exec team is arguing between €500 and €5,000 a month, a PSM study that says the acceptable range is €800-€2,400 has earned its cost by killing two positions.
Don't use it as your only method. It gives a range. It doesn't tell you where in the range to sit, and it can't see volume, margin, or competitive response.
Don't use it in an established category. If everyone knows a CRM seat costs $50-$150, respondents will parrot the market back at you and you'll have paid to confirm what's on a public pricing page. Use price benchmarking and Gabor-Granger instead.
Don't use it for a variable bill. Covered in the next section.
Van Westendorp and usage-based pricing
Here's where the 1976 design shows its age. PSM assumes the respondent is pricing a thing: one product, one price, one purchase. Usage-based pricing breaks all three assumptions, because the customer's bill depends on consumption they can't forecast, across a rate card with several dimensions.
Ask a prospect "at what price per API call would this be so expensive you wouldn't consider it" and you'll get a number. The number is noise. They don't know how many calls their workload generates, so they can't convert your unit price into a monthly cost, so they're guessing at a magnitude rather than expressing a preference.
Three adjustments that make PSM usable for consumption pricing:
Ask about the monthly bill, not the unit rate. Frame all four questions around total monthly spend at a stated volume: "for a team processing 50,000 documents a month..." Now the respondent is pricing something they can feel. Divide back to a unit rate afterward.
Run it at multiple volume points. Three studies at low, mid, and high volume give you three ranges, and the shape between them is the beginning of a tier structure. See tiered usage-based pricing.
Price the outcome, not the mechanism. For AI products, run the study on resolved tickets, enriched records, or generated documents rather than on tokens or GPU-seconds. Buyers have intuitions about outcomes and none whatsoever about tokens.
For a hybrid model with a platform fee plus overage, PSM can price the platform fee reasonably well and can't price the overage rate at all. That's a conjoint problem.
Criticisms and limitations
The OPP has no theoretical basis. It's a curve intersection that was given an attractive name. There's no economic argument that the point where "too cheap" rejections equal "too expensive" rejections maximises anything.
It measures stated preference. Nobody spends money in a survey. Stated willingness to pay runs above actual behaviour, consistently and by an amount that varies by category, so the absolute numbers deserve a haircut you'll have to size by judgment.
It's blind to volume and margin. A range with no demand curve attached can't tell you whether the top of the range costs you 5% of your buyers or 40%. The NMS extension partly fixes this.
Answers are order-dependent and wording-sensitive. Asking "too expensive" first anchors everything after it. Standard practice is to ask cheap-to-expensive or to randomise, and to keep the wording identical across waves if you want to compare over time.
Respondents who don't understand the product produce clean-looking garbage. The curves will still intersect. They'll still yield an OPP. It'll still be wrong. Screening and a clear product description do more for study quality than sample size does.
It assumes price is a proxy for value. In categories where the buyer has strong non-price preferences (security posture, vendor consolidation, an existing relationship), the price range PSM produces can be wide and uninformative because price wasn't the deciding variable.
Related terms
Frequently asked questions
What does the Van Westendorp Price Sensitivity Meter measure?
It measures the range of prices a market finds credible for a product, bounded by the point where a low price triggers quality doubts and the point where a high price triggers rejection. It measures perception of price, not quantity demanded.
What are the four Van Westendorp questions?
Too cheap (quality doubts), cheap (a bargain), expensive (would need thought), and too expensive (wouldn't consider). All four are open-ended, and no price is shown to the respondent.
What is the Optimal Price Point?
The intersection of the "too cheap" and "too expensive" curves, where equal numbers of respondents reject the price in each direction. It is not the revenue-maximising or profit-maximising price, despite the name.
What is the Indifference Price Point?
The intersection of the "cheap" and "expensive" curves, where equal numbers call the price cheap and expensive. It often approximates the perceived market price or the position of a category leader.
What is the range of acceptable pricing?
The span from the Point of Marginal Cheapness to the Point of Marginal Expensiveness. This range, rather than any single point, is the main output of a PSM study.
How many respondents does a Van Westendorp study need?
Convention is 150-300 per segment you want to read independently. Fewer than about 100 and the curves get lumpy enough that intersections shift on noise.
Can Van Westendorp give me a demand curve?
Not on its own. The Newton-Miller-Smith extension adds two purchase-intent questions and produces trial and revenue curves from them.
Is Van Westendorp better than Gabor-Granger?
They answer different questions. Van Westendorp finds a credible range when no reference price exists. Gabor-Granger finds a revenue-maximising point when the respondent already understands the category. Running PSM first and Gabor-Granger second is a reasonable sequence.
Does Van Westendorp work for B2B SaaS?
It works for the subscription component, particularly for new categories. It works poorly for the usage component, and it needs the respondent to be a budget holder rather than an end user.
Does Van Westendorp work for usage-based or AI pricing?
Only with modification. Ask about total monthly spend at a stated volume rather than about a unit rate, run it at several volume points, and frame it on an outcome the buyer understands rather than on tokens or compute time.
What's the biggest mistake when using Van Westendorp?
Shipping the OPP as the price. It's a range-finding method. Treating one intersection as a recommendation gives the number an authority the method never earned.
How is Van Westendorp analysed?
Convert each of the four answer sets into a cumulative distribution, plot all four on the same price axis, and read off the four intersections. Any spreadsheet can do it, which is part of the method's appeal.
Educational reference. Once you've picked a price, changing it should be cheap. Solvimon holds plans, rate cards, tiers, and commits as configuration, so repricing is a catalog change rather than an engineering project. See pricing methodology for the full method map.
Ready for billing v2?
Solvimon is monetization infrastructure for companies that have outgrown billing v1. One system, entire lifecycle, built by the team that did this at Adyen.
