9. When validation is done

§The question: how do we know validation is finished, and what do we do with the answer?

§Validation does not end on a date. It ends on a signal, and the signal is a combination, because any one number can lie and the three together rarely do. This chapter is about reading the combination and about the decision it forces, in both directions.

§ 9.1The three signals together#

§Retention flattens above zero. A cohort's curve levels off at some fraction that persists, and that fraction is the same or better across successive cohorts. This is the foundational signal, from chapter 6, and without it the other two do not matter.

§Activation is reliable. New users reach the first real value at a rate that holds up cohort over cohort, from chapter 5. Reliable activation means the product can turn a stranger into an active user repeatably, which is the precondition for growth spending to buy activated users rather than sign-ups.

§The target segment has paid. Revealed willingness to pay, from chapter 7, in the segment the three circles named, at a price the unit economics accept. This is the viability signal, and it is what separates a beloved product from a business.

§When all three hold, you have the honest small-scale version of product-market fit. Not a feeling, not a spike, but a container that holds users, fills reliably, and takes money from the right people.

§ 9.2The surveys, as a supporting read#

§Sean Ellis's forty-percent test, the share of users who would be very disappointed to lose the product, and Rahul Vohra's segmented version of it, are useful supporting signals, especially at a scale too small for clean retention statistics. They are a structured qualitative read, in the sense of chapter 4, and they are strongest when they agree with the behavior. A high must-have score with a decaying retention curve is a contradiction to investigate, usually meaning the enthusiasts are a smaller segment than the survey suggests. Behavior wins when they disagree.

§ 9.3The green light for growth#

§When the three signals hold, the container has a bottom, and only then is growth spending justified. The team moves to Part III. Everything before this point was about proving the product works; everything after is about getting more of what works into the hands of more of the right people. The order is not negotiable, and the discipline of the whole whitepaper is in holding it: the pressure to grow is strongest exactly before the product has earned it.

§ 9.4The other answer#

§Validation can also end by telling you the product does not work. The retention curve will not flatten despite honest experiments. Activation stays low because the value is not there. The target segment will not pay at any price the economics accept. This is not a failure of validation; it is validation succeeding at its actual job, which is to tell you the truth while there is still runway to act on it.

§The response is the fourth whitepaper's iterate-or-pivot decision, made from this evidence. A team that has done validation honestly reaches this decision with a clear grid: which of the three signals failed, and whether the failure is in the solution, the segment, or the value. Andrew Chen's framing of escaping the cold start applies in reverse here too: knowing you have not escaped it is what lets you stop rowing in the wrong direction.

§ 9.5What you leave with#

§A reading of the three signals together, by cohort. A decision, in writing, with a date: grow, because all three hold; or go back to strategy, because one or more failed, with which one named. Validation ends in that decision. If it is green, the rest of the whitepaper is yours. If it is not, the fourth whitepaper is, and you reach it with better evidence than you had the first time.