What to build first
What to build first
Version one should do the single job that justified spending the money, end to end, for one type of user. Everything else waits. The common failure is not building too little - it is building a thin slice of everything, so nothing is usable and nobody can tell whether it worked. Cut whole features, never the depth of the one feature that matters.
The cut that works
- Pick the one job that pays for the project and build it properly, all the way through.
- Serve one user type first. The second type is usually a different product wearing the same logo.
- Replace an admin screen with a spreadsheet import until volume makes it hurt.
- Ship reporting last. You will not know which numbers matter until people have used it for a month.
The cut that fails
- Half-building three features so the demo looks complete. Nobody can use any of them, so you learn nothing.
- Dropping the error handling and the boring edge cases. Those are what make software trustworthy in week three.
- Leaving out the part that saves the time, because it was the hardest. That was the reason for the project.
Common questions
- How small is too small?
- If nobody would change how they work to use it, it is too small. Version one has to be worth adopting on its own.