The Lean Startup
Eric Ries
Build it, measure it, learn from it, and do all of that before you fall in love with your own idea.
Eric Ries wrote this after watching a startup he worked on spend a long time building a product nobody wanted. His argument is that a startup is an experiment, and should be run like one. The thing you're really producing in the early days is learning about what customers actually want.
The method is a loop: build, measure, learn. Build the smallest version that can test your biggest assumption, which he calls a minimum viable product. Put it in front of real people. Measure what they actually do, not what they say they'll do. Then decide whether to keep going or change direction, which he calls a pivot. He's hard on vanity metrics, the big flattering numbers like total sign ups that make you feel good without telling you whether the product works.
A lot of it has become so standard in tech that it's easy to forget it had to be argued for. Words like MVP and pivot came into everyday use largely because of this book.
It's on the shelf because the idea runs through how I build. HackWay shipped in phases, with the core idea engine live first and the deeper tools still ahead. Its refine step even asks you to scope an MVP. The whole point is to get the smallest honest version out, and let real use tell you what to build next.
