# Van Westendorp Price Sensitivity Meter: Setup and Analysis (2026)

> How to set up and analyse a Van Westendorp price sensitivity meter: the four questions, survey setup, a worked example, R code, limits and Gabor-Granger.

Source: https://www.pulselake.co/blog/van-westendorp-price-sensitivity-meter-setup-analysis-2026
By PulseLake Team · PulseLake

The Van Westendorp Price Sensitivity Meter (PSM) is a survey method built on four price questions about one product: at what price is it too cheap, a bargain, getting expensive and too expensive? Each question becomes a cumulative curve. Where the curves cross, you get an acceptable price range plus an indifference and an "optimal" price point. It maps perception, not purchases, so it narrows the range. It won't pick the revenue-maximising price for you.

Quick disclosure: we make PulseLake, a research platform. It comes up once, near the end, and everything here works in any survey tool.

![Four crossing price curves on one chart with a shaded acceptable range between them.](https://www.pulselake.co/blog/img/production/901fb483c71fc96f05834451ba6d41727073e766-1200x840.png?w=1600&fit=max&auto=format)

## What is the Van Westendorp Price Sensitivity Meter?

It's a way of finding where price starts to feel wrong. Peter van Westendorp, a Dutch economist, introduced the technique in 1976; Wikipedia's reference list cites the proceedings of an ESOMAR congress ([Wikipedia](https://en.wikipedia.org/wiki/Van_Westendorp%27s_Price_Sensitivity_Meter); the 1976 date is also cited by the [pricesensitivitymeter package on CRAN](https://cran.r-project.org/package=pricesensitivitymeter)).

The question it avoids is "what would you pay?" People are bad at answering that. Instead it asks where price signals a problem: doubts about quality at the low end, unaffordability at the high end. Respondents name their own prices, which makes the method a good fit when you don't yet know the plausible price range for a product. If you want the shorter plain-language version first, read [Van Westendorp pricing research explained](https://www.pulselake.co/blog/van-westendorp-pricing-research).

## What are the four Van Westendorp questions?

Sources word them differently. Wikipedia gives the "too expensive" question as a price so high you wouldn't consider buying, and the "too cheap" one as a price so low you'd doubt the quality. Here's one common form:

1. At what price would you consider the product to be so inexpensive that you would doubt its quality? (**too cheap**)
1. At what price would you consider it a bargain, a great buy for the money? (**cheap** or good value)
1. At what price would it start to get expensive, though you would still consider it? (**expensive**)
1. At what price would it be so expensive that you would not consider buying it? (**too expensive**)

Every answer is typed in as a number. No pick lists. For help with wording, see [writing survey questions that don't bias responses](https://www.pulselake.co/blog/writing-questions-that-dont-bias-responses).

## How do you set up a Van Westendorp survey?

Get the setup wrong and the curves are noise, however nicely they chart. Our checklist:

1. Describe one product, once. Write a short description that stays identical across all four questions: what it is, what's included, the billing period, whether tax is included. Never change the package between questions.
1. Show no price first. Any example price in the description anchors the answers, so we leave prices out of it entirely.
1. Use numeric entry with a stated currency and period (for example "per month, in US dollars"), and validate that the field accepts numbers only.
1. Keep the order of the four questions the same for everyone. Sources disagree on the best order. Pick one, fix it, and note it in your method.
1. Screen for people who could actually buy the product, not a general panel. See [avoiding sampling bias](https://www.pulselake.co/blog/avoiding-sampling-bias). If segments matter (small versus large companies, say), plan for enough respondents in each.
1. Plan the sample. There's no standard minimum. One practical check is to run the analysis on random halves of your data and confirm the price points hold. That's our suggestion, not a published rule.
1. Put pricing after the respondent has seen the concept, and keep the survey short. See [survey design best practices](https://www.pulselake.co/blog/survey-design-best-practices-2026).

Want to try the arithmetic on a small data set? Our free [Van Westendorp calculator](https://www.pulselake.co/tools/van-westendorp-calculator) does it.

## How do you analyse Van Westendorp results?

You build four cumulative curves and read where they cross. Two of them run "downwards": the share of people whose too-cheap or bargain price is at or above a given price. The other two run "upwards": the share whose expensive or too-expensive price is at or below it. Wikipedia says two of the four curves are inverted so that the intersection points can be found.

Below are the four points, using the definitions in the [pricesensitivitymeter package documentation](https://max-alletsee.github.io/pricesensitivitymeter/reference/psm_analysis.html). One warning before you read them. Sources differ on the two marginal points: Wikipedia describes PMC as "too cheap" crossing "expensive" and PME as "too expensive" crossing "cheap". Check which convention your tool uses.

| Point | Where the curves cross | Reads as |
| --- | --- | --- |
| PMC, point of marginal cheapness | "Too cheap" meets "not cheap" | Lower bound |
| PME, point of marginal expensiveness | "Too expensive" meets "not expensive" | Upper bound |
| IPP, indifference price point | "Cheap" meets "expensive" | As many say bargain as say expensive |
| OPP, optimal price point | "Too cheap" meets "too expensive" | Fewest people rejecting at either end |

The stretch from PMC to PME is the acceptable range (the package documentation and [Wikipedia](https://en.wikipedia.org/wiki/Van_Westendorp%27s_Price_Sensitivity_Meter) both describe it this way).

## What does a worked example look like?

This one is **hypothetical**: ten invented respondents rating an invented subscription product, in dollars per month. A real study needs many more people.

| Resp. | Too cheap | Bargain | Expensive | Too expensive |
| --- | --- | --- | --- | --- |
| 1 | 10 | 20 | 35 | 50 |
| 2 | 15 | 25 | 40 | 55 |
| 3 | 15 | 30 | 40 | 60 |
| 4 | 20 | 30 | 45 | 60 |
| 5 | 20 | 35 | 45 | 65 |
| 6 | 25 | 35 | 50 | 70 |
| 7 | 25 | 40 | 55 | 70 |
| 8 | 30 | 40 | 55 | 75 |
| 9 | 30 | 45 | 60 | 80 |
| 10 | 35 | 50 | 65 | 90 |

At each price, count the share of the ten who land on each curve. "Too cheap" and "bargain" count people whose answer is at or above the price. "Expensive" and "too expensive" count people whose answer is at or below it.

| Price | Too cheap | Not cheap | Cheap | Expensive | Not expensive | Too expensive |
| --- | --- | --- | --- | --- | --- | --- |
| $30 | 30% | 20% | 80% | 0% | 100% | 0% |
| $35 | 10% | 40% | 60% | 10% | 90% | 0% |
| $40 | 0% | 60% | 40% | 30% | 70% | 0% |
| $45 | 0% | 80% | 20% | 50% | 50% | 0% |
| $50 | 0% | 90% | 10% | 60% | 40% | 10% |
| $55 | 0% | 100% | 0% | 80% | 20% | 20% |

Now read the crossings, using linear interpolation between the $5 steps:

- PMC falls between $30 (too cheap 30%, not cheap 20%) and $35 (10% against 40%). The gap goes from +10 to -30 points, so the crossing is 30 + 5 x 10/40 = **$31.25**.
- IPP is where cheap and expensive cross, between $40 (40% against 30%) and $45 (20% against 50%): 40 + 5 x 10/40 = **$41.25**.
- OPP is messier. "Too cheap" is 0% from $40 and "too expensive" stays 0% until $50, so nobody is put off at either end anywhere from $40 to $50. A calculator may report the start of that stretch ($40) or its middle ($45). The tools disagree, which is a good reason to report the stretch rather than one number.
- PME is where "too expensive" (20%) meets "not expensive" (20%), at **$55**.

That gives an acceptable range of roughly **$31 to $55**, with the indifference point near $41. Two cautions. "Optimal" means the point where rejection at both ends is lowest; it doesn't maximise revenue or profit. And ten people give coarse curves. Any point above could move several dollars with a different sample.

## How do you run it in R or a spreadsheet?

A spreadsheet is enough. Sort each question's answers, build a cumulative percentage column per curve, chart the four lines and find the crossings.

In R, the pricesensitivitymeter package (version 1.3.3, maintained by Max Alletsee) implements the analysis and a standard chart. The call below follows that package's documentation and is an **illustration**. `data_psm_demo` is a simulated data frame that the documentation's example builds itself, so swap in your own data frame and column names.

```r
library(pricesensitivitymeter)

output_psm_demo <- psm_analysis(
  toocheap = "tch",
  cheap = "ch",
  expensive = "ex",
  tooexpensive = "tex",
  data = data_psm_demo
)

summary(output_psm_demo)
```

Its documentation example reports an accepted price range, an indifference price point and an optimal price point ([package documentation](https://max-alletsee.github.io/pricesensitivitymeter/)). The package also restricts the analysis to respondents with consistent (transitive) price answers. In its simulated demo, 107 of 250 cases (43%) were removed that way. The demo data is random, so treat that as an illustration, not a typical rate. Decide your cleaning rules before you look at results, and report how many respondents you dropped. Our guide to [detecting low-quality survey responses](https://www.pulselake.co/blog/detecting-low-quality-survey-responses) covers the wider cleaning step.

## What are the limits of Van Westendorp?

- It doesn't estimate demand. Wikipedia notes that the original method addressed pricing perceptions, not demand, and that later extensions (including Newton, Miller and Smith) add purchase-likelihood questions to estimate it, which means extra questions beyond the four. Wikipedia describes the Newton, Miller and Smith extension as adding two purchase-likelihood questions to locate a revenue-maximising price, and the package documentation says its implementation estimates prices that maximise reach (trial rate) and revenue.
- It treats price as a signal of value and assumes respondents can picture the going prices in the category. For a product nobody has a reference point for, answers can be guesses.
- It ignores competitors unless you build them into the description or the analysis.
- It records stated views, not purchases.
- It covers one product at a time. If price trades off against features, use [conjoint analysis](https://www.pulselake.co/blog/conjoint-analysis-for-product-decisions) (see also our [conjoint design guide](https://www.pulselake.co/blog/conjoint-analysis-guide-design-and-analyze-2026)).

## How is price sensitivity analysis different from Van Westendorp?

"Price sensitivity analysis" is the umbrella term. It covers any way of learning how demand or perception shifts with price, and Van Westendorp is one method under it. The common approaches, by what they tell you:

| Approach | What it answers | Caveat |
| --- | --- | --- |
| Van Westendorp | What range feels acceptable? | Perception only |
| Gabor-Granger | How many would buy at each price? | Only the prices you test |
| Conjoint | How price trades off against features | More complex design |
| Past sales or experiments | What people actually did | Needs a live product and history |

Price elasticity is the usual summary number: the percentage change in quantity divided by the percentage change in price. A **hypothetical** example: raise a price from $40 to $44 (+10%) while monthly units fall from 1,000 to 950 (-5%). That's -5 / 10 = -0.5, so demand fell by half as much as price rose.

## How does Gabor-Granger work, and when do you add it?

The Gabor-Granger method, developed by Clive Granger and Andre Gabor in the 1960s, flips the approach. You choose the prices and ask about purchase intent at each. In the version Wikipedia describes, researchers pick five prices; a respondent sees one, answers on a five-point purchase-intent scale and, if they choose one of the top two levels, is shown a higher price, otherwise a lower one, until their highest acceptable price is found ([Wikipedia](https://en.wikipedia.org/wiki/Gabor%E2%80%93Granger_method)). The output is a demand curve (price against the share willing to pay) and a revenue curve.

A **hypothetical** result for five prices:

| Price | Would buy | Revenue per respondent |
| --- | --- | --- |
| $30 | 70% | 30 x 0.70 = $21.00 |
| $40 | 55% | 40 x 0.55 = $22.00 |
| $50 | 42% | 50 x 0.42 = $21.00 |
| $60 | 28% | 60 x 0.28 = $16.80 |
| $70 | 15% | 70 x 0.15 = $10.50 |

Revenue peaks at $40. Don't celebrate yet. With 200 respondents, a 55% share has a margin of error of about plus or minus 7 points (1.96 x the square root of 0.55 x 0.45 / 200 = 0.069), so $30, $40 and $50 are close to tied. Treat the result as a region to test.

Because the two methods answer different questions, teams often use both: Van Westendorp to find the plausible range, then Gabor-Granger to choose among prices inside it. Our [Gabor-Granger vs Van Westendorp answer](https://www.pulselake.co/answers/gabor-granger-vs-van-westendorp) explains the sequence and how to ask both in one survey, and the [Gabor-Granger calculator](https://www.pulselake.co/tools/gabor-granger-calculator) handles the arithmetic.

## Where does PulseLake fit?

PulseLake runs pricing surveys as traditional research, keeping the method and evidence in the same study context as the decision they feed. It doesn't set prices for you or supply market pricing data. The method above works just as well in any survey tool and a spreadsheet.

## Frequently asked questions

### What are the four Van Westendorp questions?

They ask the price at which the product feels too cheap (quality doubts), a bargain, like it's getting expensive, and too expensive to consider. Keep the product description word for word the same across all four.

### How do you read a Van Westendorp chart?

Look for where the curves cross. The lower crossing (PMC) and upper crossing (PME) bound the acceptable range. The cheap/expensive crossing is the indifference point, and the too-cheap/too-expensive crossing is the "optimal" point.

### What sample size do you need?

No standard minimum exists. Test stability by running the analysis on random halves of your data and seeing whether the price points hold, and go larger if you need to compare segments.

### Is the optimal price point the price you should charge?

Not necessarily. It only marks where rejection at both ends is lowest, and it says nothing about how many people would buy or what you'd earn.

### What is the difference between Van Westendorp and Gabor-Granger?

In Van Westendorp, respondents name the prices and you get an acceptable range. In Gabor-Granger, you show prices you chose and get a demand and revenue curve.

### Can I run Van Westendorp without special software?

Yes. A spreadsheet with cumulative percentages and a line chart does the job, or you can use R's pricesensitivitymeter package.

## Related reading

- [Van Westendorp Pricing Research Explained](https://www.pulselake.co/blog/van-westendorp-pricing-research)
- [Gabor-Granger vs Van Westendorp: which to use?](https://www.pulselake.co/answers/gabor-granger-vs-van-westendorp)
- [Conjoint Analysis for Better Product Decisions](https://www.pulselake.co/blog/conjoint-analysis-for-product-decisions)
