Founders sometimes ask us to skip discovery and go straight to a number. We understand the impulse u2014 it feels like paying for a document instead of paying for progress. But a quote produced from a paragraph and a phone call is not a commitment, it is a guess, and the person who eventually pays for the gap between the guess and reality is you.
What the week actually produces
A written definition of the problem and the specific user it is for.
A prioritized feature set, split into what launches and what waits.
The technical approach, with the constraints and risks named up front.
A fixed quote and timeline for the build, based on all of the above.
That last item is the point. Everything before it exists so the number can be real, and so both sides are agreeing to the same thing when the build starts.
The other thing it produces
Occasionally discovery ends with us telling a founder not to build yet u2014 that the problem is not sharp enough, or that a much cheaper manual test would answer the question first. That is a good outcome. It costs a week instead of a quarter, and the honesty tends to be the reason people come back.
We credit the sprint against the build if you proceed, so choosing to plan properly does not cost you anything on the way to shipping.
