In this guide 6 sections
Start with one falsifiable claim. Then interview real target users about what they did last time, and ask for a real commitment such as a deposit or a paid pilot. Set the pass marks before you begin. Polite praise is not evidence. Weak evidence means stop.
The problem and the assumption that could be wrong
A claim like “people need better project management” can never be proven false, so no interview or survey will ever settle whether the product deserves to exist. Test a sharper claim instead. Try this one: an interior design studio will pay a monthly fee to share live timelines with clients instead of scattered chat messages. That one can fail.
Write the assumption on paper before you speak to anyone, because ambiguous feedback gets read as support for whatever the team already wanted to build. A written claim keeps score. Put a date on it as well, so the team knows when the evidence is due, who gathers it and who decides.
Talking to real target users
Leading questions produce polite yes answers. “Would you use an app that saved you time on client updates?” gets agreement from almost everyone, because saving time sounds good in the abstract. Ask about the past instead. “Walk me through the last time you updated a client on project status, which tool you used, and what annoyed you about it.”
The useful interviews are with owners and project managers who can describe a real update they sent last week, including the file and the client’s reaction. Detail is the real evidence. Forecasts about the future are cheap. Record the exact words people use for their frustration, because those words later become the headline on the signup page and in the first outreach email.
Testing demand with a real cost to the user
Interviews record what people say. The next test asks for a cost: an hour of their time, a signup with a working email address, a deposit, or live use of a prototype. Charge something even for the manual pilot, since a free trial tells you only that nobody refuses free things, never that they would pay.
For the studio idea, run one real client’s project through a manual version of the concept, with a tidy weekly update document replacing the scattered chat messages. Then watch two things. Did the owner and the client both prefer the document, and would the owner pay to have it produced automatically every week? This costs a fraction of a build.
Evidence thresholds and the limits of a test
Decide what counts as strong evidence, weak evidence and a clear no before the test runs, in the same way an MVP needs its learning goals before launch. Strong evidence in the studio case is several studios asking to keep the manual version after the trial ends, or agreeing to pay before anything is automated. Weak evidence is polite interest.
Every method has a limit. A handful of interviews cannot show how sensitive buyers are to price, and a single pilot customer cannot show whether the idea holds across a whole industry. Name those limits in writing. Pretending they do not exist leads a founder to over-trust a small, encouraging signal and spend a whole build budget on it before anyone else has paid.
Proceed, refine or stop
Three next steps are responsible. Proceed to an MVP if the evidence clears the pass mark, refine and retest if it was mixed, or stop if it was weak. Mixed results are common. Often the problem is real but the proposed solution is slightly off, which is a finding and not a failure of the process.
Stopping is the outcome founders resist, because it feels like the validation failed, when it actually found a weak idea before the build budget was spent. That is the job done. Reaching the same finding after a full build would cost the team months of work, the founder’s savings and the trust of early customers.
Write the decision, the evidence behind it and the date in one paragraph, and share it with everyone who funded or advised the work.
Validation experiment worksheet
A one-page template with boxes for the falsifiable claim, the interview questions to ask, the leading questions to avoid, the commitment test and the pass marks. The last line records the outcome.
Open a blank worksheet to printQuestions owners ask
Are positive interviews enough validation?
The answer is no. Positive interviews show that people like the idea in the abstract, which is a weak signal when nothing is at stake for them. Look for time, money or habit.
How can we test willingness to pay?
Ask before the product exists. Options include a refundable deposit for early access, a signed letter of intent to pay after launch, or a paid pilot of a manual version of the service. Stated prices and committed prices rarely match.
What if the test result is unclear?
Refine one variable and test again. Change the audience, the offer or the price, but only one at a time, so the next result can be traced to a single cause. A second unclear test is a no.