Done is not the same as working
Building something that passes every internal test and shipping something a real user can actually complete are two different jobs.
The tablet had been sitting in the lobby for twelve days. Every technical check had passed. The logic was correct. The tests ran clean. And then I looked at the completion count: zero. Not low. Zero.
Nobody had made it through.
It took about an hour to find. A phone number typed without a country code - the natural way most people type it - silently matched nothing and sent the user to a dead end. There was also a screen that said "enter the 4-digit code" when there was no code to enter. Users were supposed to make one up. But the word enter implies something already exists. So they waited. One person waited twenty-two minutes for a code that was never coming.
The code was right. The logic was right. The tests were right. The experience was broken.
This happens more than founders admit. We build in a context of deep knowledge - we know what the thing is, we know what each screen means, we know which path to take. Then we hand it to someone who doesn't share any of that context, and the gap shows up immediately. Or, if nobody reports it, not at all.
The silent zero is worse than the loud complaint. A complaint tells you something. A silent zero just sits there, reading as "not enough traffic yet" or "they haven't tried it" or "it's early days." You can explain away a zero for weeks.
There's a line attributed to Patton: a good plan executed now beats a perfect plan executed later. True. But execution without observation is just motion. You have to actually watch someone use the thing. You have to look at the number and ask why it isn't higher. You have to decide that twelve days of zeros is a signal before you talk yourself into believing it means nothing.
One word fixed one of the failure points. Not "enter the 4-digit code." Make one up. Four words that changed the meaning from "wait for something" to "create something." I wrote the original version. I knew what it meant. That's exactly why I missed it.
Building for yourself is the first mistake. Shipping to yourself is the second.