Validation is a sequence, not a survey. Each stage should be cheap enough to abandon and specific enough to produce a decision.
Stage 1 - Problem
- Write the problem as a sentence naming the person, the frequency and the cost.
- Interview 15–20 people in that role. Ask what they did last time the problem occurred, not what they would do.
- Record what they currently spend in money, tools or hours to work around it.
- Decision: proceed only if a majority describe an active workaround.
Stage 2 - Customer
- Define the narrowest segment where the problem is most acute.
- Identify who holds the budget and who blocks the purchase.
- Find a repeatable way to reach 100 of them.
- Decision: proceed only if you can reliably contact the segment without paid media.
Stage 3 - Solution
- Build the smallest artefact that lets someone complete the job - a manual service counts.
- Deliver it for 5–10 users and measure whether they return unprompted.
- Decision: proceed only if there is unprompted repeat usage.
Stage 4 - Willingness to pay
- Quote a real price and ask for payment. Letters of intent are not evidence.
- Track conversion from quote to payment and the objections raised.
- Test one price point up and one down to find sensitivity.
- Decision: proceed to build only after money changes hands.
What does not count as validation
Positive survey responses, waitlist signups without payment, encouragement from people who are not the buyer, and competitor funding announcements. None of these establish that this customer will pay this price for this solution.