Sell first. Design later.

If nobody will commit before the polish, the polish was never the product. Demand is the design constraint.January 8, 2026

I used to think the right sequence was: elegant product, then customers. That sequence flatters builders. It also hides whether anyone wants the thing. The rule I ended up trusting is simpler: sell first, design later. I don't mean vaporware. I mean get a letter of understanding, a deposit, a signed intent, a LOU, a "yes, we'll buy this if it does X" — before you spend months decorating the unknown. I first felt this outside software. Construction tenders. Wind turbines. Physical demand that does not care about your Figma file. You either have a buyer on the other end of a procurement process or you have a story. Stories do not ship steel. When I brought that into software, Elite taught me the same lesson with colder numbers. Features that felt inevitable in the room died in actual sales calls. Workflows we almost overbuilt were the ones customers paid for. The market edited the product faster than we could. So the heuristic became:

Secure committed interest before you build the ceremonial version.
Let pricing and objections reshape the architecture.
Treat "we need just one more polish pass" as a delay strategy unless a specific buyer named the polish.

This does not mean shipping garbage forever. It means the first artifact that matters is not the homepage. It is the agreement that someone will use the thing when it works. YC companies get romanticized as product geniuses. A lot of them are just unusually good at finding demand before they fall in love with their own taste. If you cannot get a yes before the redesign, the redesign was probably not the missing piece.