Every founder I talk to has a number in their head. Twenty-nine a month. Ninety-nine a seat. Five grand for the pilot. Ask where the number came from and you get some version of "it felt about right" or "that's what the competitor charges."
That's not pricing. That's a guess wearing a dollar sign.
Willingness to pay is the amount a specific customer will actually hand over for a specific outcome, and it's the one validation signal that can't be talked around. Someone can love your demo, forward it to a colleague, join your waitlist, and still never pay you a cent. The gap between enthusiasm and payment is where most first-time founders lose six months. So let's close it before you write the code.
What does willingness to pay actually mean before you have a product?
Willingness to pay is the highest price a buyer would accept before they walk away, measured by behaviour rather than opinion. Pre-launch, you're not measuring a number. You're measuring whether a number exists at all.
There's a distinction that matters here and almost nobody makes it. There are two separate questions hiding inside "will they pay?":
- Will anyone pay anything? (Does the problem hurt enough to open a wallet?)
- How much will they pay? (Where does the number land?)
Question one is a validation question. Question two is a pricing question. If you try to answer both with the same test, you get a muddy result and usually an optimistic one. Answer question one first. A founder who knows eleven people will pay something has a business. A founder who knows the theoretically optimal price for a product nobody wants has a spreadsheet.
Why do price surveys give you the wrong number?
Because people are bad at predicting their own future spending, and worse at disappointing you to your face. Hypothetical questions produce hypothetical answers, and hypothetical answers skew high on interest and low on price.
Ask "would you pay $40 a month for this?" and you're asking two things at once: do you like me, and is $40 reasonable in the abstract. Most respondents answer the first question. Research on the Van Westendorp method, which I'll get to in a second, is explicit about this: results are biased by the hypothetical framing and by the method's focus on the point of minimum customer resistance rather than maximum revenue.
That doesn't make surveys useless. It makes them a range-finder, not a decision. Here's the hierarchy I'd use, weakest signal at the top:
| Signal | What it proves | Strength |
|---|---|---|
| "Yes, I'd pay for that" | They're being polite | Near zero |
| Waitlist signup | Mild curiosity, zero cost to them | Weak |
| Survey price ranges (Van Westendorp) | Rough price corridor | Directional |
| Pricing page click to checkout | Interest survives a real number | Moderate |
| Refundable deposit | Money moved | Strong |
| Signed LOI or paid pilot | Budget owner committed | Strongest |
Notice the pattern. Signal strength tracks exactly with how much it costs the respondent to say yes. Free yes, worthless. Costly yes, meaningful. That's the whole framework, and if you remember nothing else from this article, remember that line.
We wrote a longer piece on why waitlists specifically fail this test, and the short version is that a waitlist measures curiosity at a price of zero, which tells you almost nothing about behaviour at a price above zero.
How do you run a Van Westendorp price sensitivity test?
You ask four questions about price after showing a clear product description, then plot the answers to find the range where most people think price and value line up. It takes about 40 respondents to be useful and about 200 to be stable.
The four questions, in this exact order:
- At what price would this be so expensive you wouldn't consider it?
- At what price would this be expensive, but still worth considering?
- At what price would this feel like a bargain?
- At what price would this be so cheap you'd question the quality?
Plot the cumulative curves and you get two useful intersections. The point where "too cheap" crosses "too expensive" is the price of indifference. The range between the other two crossings is your acceptable corridor. Tools like Conjointly, Qualtrics, and SurveyMonkey will draw the chart for you, or you can do it in a spreadsheet in twenty minutes.
Three rules for making it worth the effort:
Describe the outcome, not the features. "A tool that cuts your monthly close from five days to one" gets you a real answer. "An AI-powered financial workspace" gets you noise.
Recruit the actual buyer. Twelve responses from people who have the problem and the budget beat 300 from a general audience. If you're B2B, the person who'll sign the invoice is the only respondent who counts.
Treat the output as a corridor, not a price. Van Westendorp tells you where the walls are. It does not tell you where to stand.
What's the strongest signal you can get before you build?
A payment. Failing that, a signed commitment from someone with budget authority. In B2B, a paid pilot or a signed letter of intent is the strongest demand signal that exists, and it's the one investors actually weigh.
The ladder, from easiest to run to most convincing:
Refundable deposit. Ask for $20 or $50 against a future subscription, fully refundable, no questions. You're not raising money. You're buying information. The friction of entering a card is the test, and a deposit conversion rate above roughly 1.5% of qualified traffic is a real proceed signal.
Pre-order at a founding price. Charge for annual access at a discount before launch, with a hard refund promise and a delivery date you'll actually hit. Fieldboom famously got 100 paying customers before writing a line of code. That's an outlier, but the mechanic is boring and repeatable.
Paid pilot. B2B only. You charge a real fee, usually a few thousand, for a scoped engagement that you may deliver semi-manually. If a company pays for a pilot, the problem has a budget line. Three to five signed pilots or LOIs is roughly the bar most seed investors treat as convincing.
Letter of intent. Non-binding, but it forces a named person to put the commitment in writing, which surfaces every hidden approval step in their org. The LOI that dies in procurement taught you something valuable and cost you a week.
One warning on pre-selling. Take money and you've made a promise. Set a date you can defend, write the refund terms in plain English, and refund without argument if you slip. A founder who refunds cleanly keeps the relationship. A founder who goes quiet burns the only twenty people who believed them early.
How do you test willingness to pay with a pricing page?
Put up a real pricing page with real numbers and a real checkout button, drive qualified traffic to it, and measure how many people click through to pay. This is a smoke test, and it's the cheapest quantitative read you'll get.
The setup takes an afternoon:
- One landing page that describes the outcome, with three price tiers.
- A checkout button that leads to either a real payment flow or an honest "we're onboarding in batches, leave your email and a deposit to hold your spot."
- Traffic from somewhere qualified: a niche subreddit, a LinkedIn post to your own network, a small paid test, a relevant newsletter.
- Enough volume to mean something. Under 200 qualified visitors, you're reading tea leaves.
Then run the same page at two different price points. Split traffic, or run price A for a week and price B the next. What you're looking for isn't just which converts better. It's whether the conversion rate moves at all. If doubling the price barely dents conversion, you're underpriced and you've learned something worth more than the test cost. If halving it doesn't lift conversion, price isn't your problem. The product promise is.
Be careful with the fake-door version of this. Taking someone to a dead end feels clever and costs you trust. An honest pre-launch checkout with a deposit gets you a stronger signal and leaves the relationship intact.
What numbers count as a pass?
There's no universal threshold, but there are ranges that experienced founders use as a gut check. Here's what I'd treat as a proceed signal at pre-launch scale:
| Test | Weak | Worth continuing | Strong |
|---|---|---|---|
| Pricing page visit to checkout click | Under 2% | 3 to 5% | Over 8% |
| Qualified traffic to paid deposit | Under 0.5% | 1.5% | Over 3% |
| Customer conversations to paid pilot (B2B) | 1 in 40 | 1 in 15 | 1 in 8 |
| Pre-orders from your own network of 100 | Under 3 | 6 to 10 | Over 15 |
Two caveats. These assume qualified traffic, meaning people with the problem, not your Twitter followers being supportive. And the B2B numbers assume you're talking to budget holders, not enthusiastic managers who then need to ask someone.
If you land in the weak column, that isn't a failure. It's a finding, delivered three months before you'd have got it the expensive way. The usual next move is to narrow the audience rather than drop the price. Weak willingness to pay across a broad group often hides strong willingness to pay inside a narrow one.
How do you turn a tested number into an actual price?
Combine the corridor from your survey with the behaviour from your paid tests, then set the launch price near the top of what behaviour supports, not the middle of what the survey suggested. You can raise prices later, but the first hundred customers anchor your positioning for years.
A worked example. Say your Van Westendorp corridor lands between $19 and $65, with a price of indifference around $34. Your pricing page test shows $29 and $49 converting within a point of each other. Your five paid pilots came in at $2,000 each without a negotiation. What that pattern says: buyers aren't price sensitive in this band, and $34 was the survey being polite. Launch at $49, and watch churn and objections for the first sixty days.
This is the point where the numbers need to live somewhere other than your head. You've got a price corridor, conversion data at two points, pilot revenue, and a cost base that determines whether any of it works. Most founders keep this in four different places and lose the thread. A spreadsheet is fine. Notion is fine. A planning tool like Foundra walks first-time founders through the pricing and unit economics sections in one place if a blank sheet is the thing stopping you. The tool matters less than the discipline of writing the assumptions down where you can check them against reality in ninety days.
What mistakes kill these tests?
Four show up again and again:
Testing on friends. Your network will pre-order out of affection. Weight those sales at roughly a tenth of a stranger's.
Asking about price before establishing the problem. If the respondent doesn't feel the pain, every price is too high, and your data is about them, not your product.
Running one test and calling it validated. One signal is an anecdote. A survey corridor plus a pricing page test plus three paid commitments is a case.
Confusing "no" with "not yet." Timing kills more deals than price. When someone declines, ask what would have to be true for this to be worth paying for. The answer is usually your roadmap.
Refusing to test high. Almost every first-time founder underprices. Test a number that makes you slightly uncomfortable to say out loud. Worst case, nobody clicks and you've bought that information for the cost of a landing page.
Key takeaways
- Willingness to pay is measured by behaviour, not opinion. The strength of any signal equals what it cost the person to give it.
- Answer "will anyone pay anything?" before "how much?" They're different questions and mixing them produces mush.
- Van Westendorp gives you a corridor in an afternoon, but its results skew low and hypothetical. Use it to frame tests, not to set price.
- Refundable deposits, pre-orders, paid pilots, and signed LOIs are the signals that hold up. Three to five paid pilots or LOIs is the rough bar for B2B credibility.
- Run a pricing page smoke test at two price points. If conversion barely moves when you double the price, you're underpriced.
- Launch near the top of what behaviour supports. Your first hundred customers anchor your positioning.
FAQ
How many people do I need for a willingness to pay test?
For a Van Westendorp survey, 40 qualified respondents gives you a usable shape and 200 gives you a stable one. For behavioural tests, you need around 200 qualified visitors to read a conversion rate, or 5 to 10 real sales conversations for B2B.
Can I test willingness to pay without a product?
Yes. A landing page with a clear outcome promise, real prices, and a deposit or pre-order flow tests it without any product existing. That's the entire point of running the test before you build.
Is it ethical to charge before the product exists?
It is, if you're honest about the delivery date, offer a clean refund, and actually refund when you slip. Pre-orders are a normal commercial arrangement. Silence after taking money is not.
What if people say yes but never pay?
That's the expected result and the reason you're testing. Verbal yes converts to payment at a low single-digit rate for most pre-launch products. Stop counting verbal commitments as pipeline and the picture gets clear quickly.
Should I test price or product first?
Together. Price is part of the offer, and an offer tested at zero cost tells you nothing about the offer at $49. Put the number on the page from the first test.
What if my willingness to pay comes back too low to build a business?
Narrow the audience before you drop the price. A segment with acute pain and budget will often pay five times what a general audience will for the same product. If no segment clears your cost base, you've saved yourself a year, and that's a win worth taking.
Top comments (0)