Minimum viable product is one phrase hiding three very different bets. Building the wrong one wastes months, so before you scope anything, decide which question you are actually trying to answer.
Prototype: answer a design question
A prototype exists to test whether an idea makes sense to a human. It can be throwaway, faked at the edges, and unconnected to a real backend. If you are still arguing about the shape of the thing, build this first.
Pilot: answer a value question
A pilot goes to a small set of real users doing real work. It has to be trustworthy where it counts and can be rough everywhere else. The goal is a clear signal that people get value, not a polished launch.
- A handful of committed early users
- Real data on the paths that matter
- Permission to be ugly outside the core
Do not launch to answer a question a pilot could have settled for a tenth of the cost.
Launch: answer a scale question
A launch assumes the value is proven and now bets on reach. This is where reliability, onboarding, and the unglamorous edges finally earn the investment — because now everyone sees them.
