From Problem to First MVP: What to Build and What to Postpone
Before you open Figma or write a line of code: how to decide on the smallest version that truly tests the idea.
The most dangerous moment in any new project is the excitement, when you want to build everything at once. A real MVP is not a small version of the product. It is the smallest possible experiment that proves or kills the hypothesis the whole project stands on.
Find the hypothesis that sinks the project if it fails
In the gift project, the hypothesis is not that I can build a recommendation engine. That is just a matter of time. The dangerous hypothesis is that people will trust a recommendation from a system and pay for it. Test that one first.
Build only what tests it, postpone the rest
A landing page explaining the promise + a form collecting details about the person + manual recommendations behind the scenes = a complete MVP. No AI, no app, no dashboard. If people pay, we build. If not, we saved months.
The two-week rule
If the first testable version needs more than two weeks, something in the scope is wrong. Cut again.