Skip to content

Price and willingness to pay

How do I find out if people will actually pay?

Updated 19 August 2026 · Thresholds read live from the verdict engine

Short answer

Stop asking and set up a moment where saying yes costs something. Take a refundable pre-order, store a card without charging it, or sell a paid pilot by hand. Then turn the rate into money: divide what you pay for a click by your conversion rate and you have the cost of a customer, which is the number that decides whether the business works.

Why asking produces a useless answer

People are bad at predicting their own future spending and good at being kind to someone who is clearly invested. Both effects push the same way. Survey answers about willingness to pay run consistently above what the same people do when a card is in front of them.

Pricing research methods like Van Westendorp and Gabor-Granger have their place in established categories, where you are moving an existing price rather than discovering one. For an unbuilt app they inherit the same flaw: the respondent pays nothing to answer, so the answer costs them nothing to get wrong.

The commitment ladder

Every rung costs the person more and tells you more. Climb until the answer changes your decision.

SignalWhat it costs themHow much it predicts revenue
Says they like itNothingNothing
Gives an emailA little attentionWeak
Answers a five-question quizTwo minutesWeak, but it filters
Reaches a paywall and picks a planA decisionModerate
Enters card details, no chargeTrustStrong
Pays a refundable depositMoney, brieflyVery strong
Pays for a pilotMoneyAs strong as it gets pre-build

Three ways to collect a paying signal

1. Card stored, nothing charged

A Stripe setup intent collects and stores card details without taking money. Say so on the screen in plain words, and again on the confirmation. This is the highest-volume paying signal available before a product exists, because it can run at the end of an ad-driven funnel and be measured against benchmarks.

2. A refundable pre-order

Take the money, state the launch window, refund every buyer the day you decide against building. Stronger than a stored card and much harder to run at volume: expect chargeback risk if you go quiet, and a support load you will have to carry personally.

3. A paid pilot, sold by hand

For B2B and higher prices, forget the funnel. Send 40 emails or 20 direct messages offering a pilot at a real price, and count how many say yes on a call. Three paying pilots teach you more than a thousand visitors ever will, and they come with people who will tell you what to build.

  • Charge what you plan to charge. A discounted test measures the discount.
  • Put the terms on the screen where the decision happens. Not in a footer, not in an email afterwards.
  • Keep the refund promise trivially easy. One reply, no form. The people who trusted you deserve that, and your ad account depends on it.
  • Record which plan they picked, and not only that they paid. Plan choice tells you about price shape.

Turning a conversion rate into money

A conversion rate is only useful once you have converted it into a cost per customer. The arithmetic is one division.

Cost per customer = cost per click ÷ conversion rate.

Cost per clickConversionCost per customerVerdict at $60 a year
$0.803.0%$27Workable, pays back inside the first year
$1.202.0%$60Break-even in year one, needs renewals
$1.201.0%$120Underwater unless people stay two years
$2.504.0%$63Expensive clicks, saved by a strong funnel
$0.500.8%$63Cheap clicks that were not really buyers
Cost per customer here means cost per checkout at test time. Live campaigns run worse as the audience saturates, so treat these as optimistic.

The last two rows are the ones people miss. A cheap click and a bad rate cost the same as an expensive click and a good rate, and the expensive one has a better business behind it. This is also why testing in a market you would never sell to flatters the top of the funnel and lies about the bottom.

Worked example

When a passing test still says no

A meditation app tested at $29.99 a year and converted 2.4% of 500 visitors, which clears the build line. Clicks cost $1.90 in the US, so each checkout cost about $79.

At $29.99 a year with a plausible two-year life, a customer is worth about $60 gross. The idea passed the demand test and failed the arithmetic. Their options were a higher price, a cheaper channel, or a different audience, and they took the first: retested at $59.99, converted 1.9%, and the same click cost produced a customer at $100 against $120 of value. Marginal, and honest.

The numbers

The bands the engine applies, and what they cost you at typical click prices.

Funnel starts before any verdict
30
Below this, don't build
1.0%
At or above this, build
2.0%
Visitors before a segment counts
50
Don't build
Promising
Build
0.0%1.0%2.0%5.0%+
Share of cold visitors who reached for a card. Under 1.0% the idea is dead at this price and angle, 1.0% to 2.0% is within optimisation reach, and 2.0% is where shipped web2app funnels already sit. The scale tops out at 5%, the top of that range.

Under 1% a customer costs more than $100 at a $1 click, which almost no consumer app price supports. At 2% the same click produces a customer for $50, which most annual plans above $50 can carry. The thresholds are demand thresholds and unit-economics thresholds at the same time, which is why they sit where they do.

ResultRate95% rangeConfidenceReads as
12 / 5002.4%, and the range clears the line2.40%1.38% to 4.15%mediumBuild
5 / 5001.0%, undecided in both directions1.00%0.43% to 2.32%mediumPromising
3 / 6000.5%, decided against0.50%0.17% to 1.46%mediumDon't build
Wilson 95% intervals from the engine. Before you act on a rate, check that the range beneath it agrees with you.

How VerifyToLaunch does this

The funnel ends in a checkout that carries your real price and real plans. The visitor picks one and confirms, and that confirmation is what the verdict counts. No card is asked for and nobody is charged, which the screen says plainly, so there is no payment account to connect before you can run a test.

We do not do the unit-economics arithmetic for you, because we do not know your click costs or your retention. The dashboard reports spend if you enter it, and the division on this page is yours to run. A verdict of BUILD IT is a statement about demand, and it is possible to clear it and still have a business that cannot pay for its own traffic.

Common questions

Is it legal to collect card details without charging?
Yes, through a payment processor's setup intent flow, which is designed for saving a payment method for later use. Disclose that no charge is being made, keep the card data with the processor rather than on your server, and honour any request to delete it.
How many buyers do I need to prove people will pay?
The count matters less than the rate and the traffic behind it. Ten checkouts from 300 cold visitors is a strong signal; ten from 3,000 is a clear no. Our engine will not issue a verdict at all under 30 funnel starts.
What if people say they would pay but the funnel converts at zero?
Believe the funnel. The gap is usually one of three things: the audience you bought is not the audience you interviewed, the promise on the page is weaker than the one you say out loud, or the price is above what strangers will risk on an unknown brand.
Should I test a lower price to get a signal?
Only if you would sell at that price. A test at $9 that passes tells you nothing about a product you intend to sell at $60, and it is a common way to talk yourself into building the wrong thing.

Put your price in front of strangers

The generated funnel carries your real price and ends in a checkout where a stranger picks a plan and confirms, so the signal you collect is commitment rather than curiosity.

Start a test

One-time licence, 30-day money-back guarantee