01
Stage one: validate the problem before the product
Start with conversations, not code. Define one specific person with one painful, recurring problem, then interview at least ten of them about how they handle it today. Listen for money, time or workarounds already being spent. Worked example: a founder imagines a scheduling tool for dog groomers, but interviews reveal that groomers mostly lose income from last-minute cancellations. The problem shifts from calendars to deposits and reminders. Write down the sentence that summarizes the problem in the customers' own words, and treat it as a hypothesis that later stages must confirm or overturn.
02
Stage two: shrink the offer to one testable promise
With a validated problem, choose the smallest offer that could solve it and describe it on a single page with a price or clear commitment. Continuing the example, the founder offers a simple deposit-and-reminder text service for five local groomers, handled partly by hand. Decide what signal counts as success in advance, such as three groomers agreeing to pay for a month. Resist adding features. The purpose of this stage is learning whether people will commit, not producing a polished product, so manual work behind the scenes is acceptable and often more informative.
03
Stage three: set the launch and review what happened
Pick a launch date, a small audience and a short list of measures you will examine afterward. In the example, the founder opens the service to ten groomers within two weeks and tracks sign-ups, no-show reductions reported by the groomers, and reasons for declining. After the window closes, hold a review: keep, change or stop. Record what surprised you and choose the next experiment. A first launch is successful when it produces a clear decision backed by evidence, even if the decision is to adjust the offer or return to interviews.