Launch to learn, not to impress
The first release isn't your product's big moment. It's the first honest answer the market gives you.
There's a way of launching an app that looks a lot like a wedding: months of preparation, one big day, everyone smiling, and an unspoken agreement not to ask hard questions. The morning after, the metrics dashboard is quiet in a way nobody planned for, and the team quietly starts calling version 1.1 'the real launch.'
The launches that go somewhere treat day one differently — not as a verdict on the team, but as a question put to the market. You don't launch to be told you're great. You launch to find out what's true.
Decide what needs to be true
Every product rests on a short stack of assumptions. People will grant this permission. People will pay before that feature. People will come back on day seven without being bribed. Most of the stack can be wrong and you'll survive — but one or two of those assumptions are load-bearing. Name them before you ship, out loud, in writing. The first release exists to test the load-bearing ones and almost nothing else.
Build the smallest thing that can be wrong
MVPs have a bad reputation because most of them are just small, and small isn't the point. Falsifiable is the point. A first release should be polished exactly where trust is earned — onboarding, the first five minutes, the moment money changes hands — and unapologetically plain everywhere your test doesn't reach. Nobody churns because settings looked basic. They churn because minute one didn't convince them.
Pick your numbers before, not after
Decide what 'working' means before launch day: two or three numbers, written down, with honest thresholds. Retention over downloads, almost always — downloads measure your marketing, retention measures your product. If you define success after the data comes in, everything will look like success, and you'll learn nothing at the exact moment learning is cheapest.
A quiet launch that teaches you something real beats a loud one that flatters you. The market's first answer is rarely yes — but it's almost always useful, and useful is what version two is made of.