We've designed dozens of MVPs, and the post-mortems of the failed ones rhyme. Here are the five mistakes we see most — and how to dodge them.
One: building for every user at once. An MVP exists to delight a painfully specific person. If your first version serves 'everyone aged 18–55', it serves no one. Pick one user, one core job, and nail it.
Two: shipping twenty average screens instead of eight excellent ones. Every extra screen dilutes your focus, your budget and your launch date. Cut until it hurts, then cut once more.
Three: skipping the prototype. A clickable prototype costs days and saves months. Watching five real users tap through it will teach you more than fifty stakeholder meetings.
Four: treating design as decoration. Design is how your product works, not how it's painted at the end. When design leads, engineering builds the right thing once instead of the wrong thing twice.
Five: launching without a feedback loop. Analytics, crash reporting and a way to talk to users are not 'phase two'. They are the whole point of shipping an MVP — learning fast.
Avoid these five and your MVP does its real job: proving the idea with the least money and the most signal.