No Single Question Will Tell You the Ideal Price
But with a learning mindset, each cycle of research can build confidence.
In our last post, we shared that 82% of UK adults want everyone to pay the same price. Read it here. One honest price. No dynamioc pricing games. Which means you had better pick the right one. Easy with slews of data from established products, but difficult for a new to the world innovation that has never been sold before.
Price too high, and a great idea dies on the shelf. Price too low, and you give away the money you need to invent the next one. So you do what every smart team does. You ask customers:
"How much would you pay?"
Every pricing question is flawed.
Ask any experienced researcher and they will list the problems. The question is too optimistic. It is too hard to understand. It is biased by the questions that came before it. I could go on and on.
They are not wrong. A new working paper from researchers at Bocconi University, Stanford, and NYU Stern proves it. Across 12 main studies and 9 supplemental studies, with 12,883 participants, two respected ways of asking "what would you pay?" gave substantially different answers. Same product. Different question. Different number.
Average willingness to pay, by how the question was asked
Bars are scaled within each product.
Even a $20 gift card, whose value everyone knows, was worth $17.77 under one question and $15.42 under the other.
Flawed but still useful.
When you want to build confidence and reduce risk, you have to buy into a learning mindset. No one question produces the perfect price. But with enough cycles of learning, you can build real confidence in the number you launch with.
Three habits that build confidence.
Ask the same way every time.
A person with one watch always knows the time. A person with two watches is never sure. If you always ask "How much are you willing to pay?" as an open-ended question, your watch may run fast. Teams in the same category often develop rules of thumb, like knocking a set percent off the answer. The research does not support a set discount, but you work with what you have. Staying consistent gives you more chances to learn over time.
The trap: if you have only ever tested non-innovative offerings, your watch has never met a breakthrough.
Triangulate with smaller samples.
Do not bet your whole budget on one big survey with one question. Run smaller studies that try more than one approach, then look for the value where they agree. Two watches can leave you unsure, but two methods that agree build confidence. Disagreement teaches you something.
Our approach maximizes learning with each research cycle.
Go head to head.
Use paired comparison surveys to see at which price point you can beat the competition, and why. You get a decision and an explanation in one study. Here is how paired comparison surveys work.
Where the usual questions break.
The same study tested a fifth product alongside the chocolate, the massage, the cleaning service, and the gift card: a pill that cuts your sleep requirement in half. No such pill exists. And it was the one product where the two questions did not produce a statistically significant difference.
Because the answers were all over the map. We know that uniqueness causes variance in purchase intent.
average answer when people typed a number.
average answer from the price list.
spread in answers (standard deviation) under both methods.
Answers could range from $0 to $2,000, and most landed at the low end. A rank-based test did detect a difference, but the averages were unstable.
Why? Big promise. No proof.
Cutting your sleep in half is a wow promise, and a truly unique one. People would pay a lot for it. But it lacks proof. There is not enough reason to believe it works.
So people split into two camps. Some assume it works, or will, and answer high. Others decide it is a gimmick or a scam and answer very low. Average the two camps together and you get a number nobody actually holds.
The more unique the idea, the less precision you can expect from "what would you pay?"
Lead the customer, and have them react.
Stop asking people to invent a price for something they have never seen. Give them one and watch what happens. Reacting to a price is something customers do every day. Inventing one is not.
Set a deliberately high price.
Purposefully charge 50% more than the best alternative.
See where purchase intent lands.
Have customers rate their purchase intent and how unique they find the idea.
Score the value.
Then have them rate the price itself. That tells you whether customers see the price as fair for what they get.
If intent holds up at a premium, you have evidence to charge more. If it does not, you found out before you launched, and you can lower the price or strengthen the proof.
Price is not a number you find. It is confidence interval you discover, one research cycle at a time.
Sources
- Gao, R. Y., Huang, S. H., and Jung, M. H., "Theoretically Equivalent Elicitation Methods Produce Systematically Different Valuations," working paper, March 2026 version (SSRN 4484841). The authors ask to be contacted for the current version before citing.
- IT Brief UK, on the HyperFinity survey of 2,000 UK adults, April 29, 2026