Van Westendorp

The obvious way to set a price is to ask customers what they would pay. The answers are close to useless. The person answering is the one who would hand over the money, so there is a quiet incentive to name a low number, and most people genuinely don't know their own ceiling until a real invoice is in front of them.

Pagedeck ran into this while pricing its team plan. Two hundred people on the free plan were asked what they would pay per month and the median answer came back at $2, while a competing tool solving the same problem charged $9. Nobody had lied. The question had asked for something people cannot answer.

Van Westendorp changes what gets asked. Instead of a fair price, it asks for boundaries. People can rarely name the right number, but they answer "this is so expensive I'm out" and "this is so cheap something must be wrong with it" with surprising consistency. The Dutch economist Van Westendorp published the approach in 1976 and the name stuck to it.

Ask four boundary questions in a fixed order, turn each set of answers into a cumulative curve, and read the price range off the points where those curves cross.

1. Pick the respondents. Only people who could plausibly buy the product belong in the sample, and 150 to 200 answers is the usual floor. Let people outside the target market in and you are averaging in prices from people who were never going to buy, which drags the whole range down.
2. Describe the product in one paragraph so everyone is pricing the same thing. Answers move by a factor of two depending on how much of the feature set the description mentions, so use the wording you would actually sell with.
3. Ask the four questions in order: at what price does this start to feel cheap, at what price is it so cheap you'd doubt the quality, at what price does it start to feel expensive, and at what price is it so expensive you wouldn't buy. Leave them open — no multiple choice, just a number typed in.
4. Drop the answers that contradict themselves. The four numbers should climb: too cheap, cheap, expensive, too expensive. Somebody who writes $4 for expensive and $7 for cheap misread the questions and comes out of the dataset.
5. Walk up the price axis and compute four cumulative percentages at each step. At the $8 mark you record the share who called that price expensive, the share who called it too expensive, the share who still see it as cheap, and the share who find it suspiciously cheap.
6. Read the crossings. Where too cheap and too expensive meet is the price that draws the fewest refusals, and the two crossings on either side of it mark the bottom and the top of the acceptable range.

Pagedeck's run came back with a range of $6 to $14 and the fewest-refusals point at $8. That is where the survey's job ends; picking a spot inside the range takes other evidence.

The method earns its keep when there is no reference price at all — a first product with nothing comparable to line up against, or a feature that has only ever been free and now needs a starting number. It carries much less weight when the question is whether to raise the price of something already selling. Purchase records beat stated intentions, and you already have them.

There are four limits worth knowing before you run it.

1. Stated prices and paid prices differ. Nobody's card is charged in a survey, so the upper boundary comes back more generous than reality.
2. It says nothing about volume. You learn the range but not how many people buy at each point inside it, which means the revenue-maximising price is not something this survey can find.
3. Ask someone unfamiliar with the category and the numbers come from nowhere. A person who has never used a collaborative document tool has no basis for weighing $2 a month against $23.
4. Products with several plans can't be covered in one pass. A free, team and enterprise structure needs a separate run per plan, and at that point price anchoring — placing the numbers side by side so each one is read against the others — moves buyers more than the survey range does.

So treat the output as a narrowing device and make the decision by stacking other evidence on top of it. The floor comes from cost-plus pricing, which adds a margin on top of what delivery costs: price below what one account costs to run and every sale loses money. The ceiling comes from value-based pricing, which converts the hours a customer saves into money. And whether the number you picked supports a business is settled by unit economics — what one customer pays you against what that customer costs to acquire.

"The Van Westendorp run put the acceptable range between $6 and $14."

A band with a floor and a ceiling now exists, and the choice happens inside it.

"Only 40 people answered, so I wouldn't take that range at face value."

The sample is small enough that the crossings shift on a handful of responses.

"The fewest-refusals point is $8, but I want to go out at $11."

Read the survey as a range rather than an answer, and pick a higher spot inside it.

Mia ran this survey before Pagedeck reworked its team plan. The four questions went to 180 account owners who had been on the free plan for three months or more, and twelve self-contradicting responses came out, leaving 168 to draw the curves from. When Grace saw the result, her first question was about the ceiling rather than the optimum. Once the top of the range is known, how far below it to sit becomes a question about revenue targets rather than a question about customers.

The most common misreading treats the fewest-refusals crossing as the revenue-maximising price. Of the crossings on the chart, it's the one that looks most authoritative, so the mistake is an easy one to make. But that point only marks where the fewest people say no, and it holds no information about how many people say yes. Used as-is, it tends to land below the middle of the range, and the money that could have been charged is gone before the plan launches.

The second habit is telling respondents what you charge, or what a competitor charges, on the theory that people need a reference point to answer sensibly. What happens instead is that answers cluster around whatever number they saw first. Show $9 and all four answers arrive close to it. The range you get back is the number you supplied, handed to you again, and you finish without learning whether there was room above it.