I know this because I went back and checked. The eight rounds of questions, the verdict logic, the no account, no cloud, nothing-leaves-your-phone part, all of that was done in July. What I spent six weeks on afterwards was wording. The copy on the fourth onboarding screen. Whether the Kill verdict's colour should read as harsh. Screenshot captions. A FAQ answer I rewrote so many times I lost count of which version was actually live.
None of it was nothing. Wording matters. But none of it was the reason the app wasn't out yet, and I knew that the whole time, which is the part that actually bothers me looking back.
The eleven versions of the same paragraph
At some point I opened the Notes app on my phone and found eleven drafts of the fourth onboarding screen, the one that explains why there's no sign up and nothing is sent anywhere. I read all eleven in order. The eleventh was, almost word for word, the same as the fourth. I had spent real hours getting back to roughly where I'd already been.
That's the tell I now watch for, in myself and in friends who ask me when they should ship. If you're rewriting something and the newest version keeps circling back to an earlier one, you've stopped editing and started stalling. The work is finished. You're just not ready to find out what happens next.
What I was actually afraid of
Not bugs. I'd run the eight rounds of questions against every feature idea I could think of, including some deliberately stupid ones, and the verdicts came back consistent every time. That part I trusted.
What I was afraid of was something closer to being read. Once it's on the App Store, a stranger opens it, answers eight honest questions about something they actually care about, and gets a verdict from logic I wrote. If the verdict feels wrong to them, that's not a bug report, that's a judgment on whether I understood the thing I was trying to build. Wording the onboarding screen for the twelfth time felt like doing something. It wasn't. It was choosing a smaller, softer fear over a bigger one I hadn't named yet.
The question that actually moved it
What got me to submit wasn't a deadline, or a checklist, or anyone telling me it was good enough. It was one question I made myself answer honestly: if a stranger opened exactly what is on my phone right now, today, and paid 4.99 euros for it, would I be comfortable standing next to them while they used it?
The answer was yes. It had been yes for weeks. I just hadn't asked the question in those terms before, because "is it good" and "is it ready" turned out to be two different questions, and I'd been answering the first one over and over while the second one sat there unanswered.
The delays that were actually real
I want to be fair to the six weeks, because not every day in them was vanity. Two things I found and fixed in that stretch were genuinely load bearing. One was a path through the question flow where answering "not sure" three times in a row produced a Build verdict that made no sense, a real logic gap, not a wording problem. The other was that the app's dark mode made the Prove verdict's accent colour nearly unreadable against its own background, which I only caught because I happened to use it outside at night.
Both of those were correctly worth delaying for. I can tell they were different from the onboarding paragraph because fixing them changed what the app did, not just how it sounded. That's the actual line for me now. A change that alters behaviour is readiness work. A change that only alters phrasing, after the second or third pass, is usually fear wearing a work costume.
What I'd tell someone stuck on this now
Three questions, and they're blunter than the ones I actually asked myself in July, because I didn't have them yet. Does the core thing work on every path you can think of to break it, not just the obvious one? Have you used it yourself, on a real decision you actually had, and trusted what it told you? And if you handed exactly what exists right now to fifty strangers today, is the thing you're afraid of a flaw they'd find, or just the fact that they'd see it at all?
If the first two are yes and the third is "they'd just see it," you are not improving your MVP anymore. You're rewriting the fourth screen for the twelfth time. Ship it, and go find out what the real feedback actually is, because you already know the imagined kind by heart.
Appray walks you through the same territory in eight rounds of questions, then puts every feature you are considering on trial against what you actually said. The result is a Build, Kill or Prove verdict for each one, plus interview questions written from your own answers rather than a template.
It runs entirely on your iPhone or iPad. No account, no sign in, nothing leaves your device, and no AI is involved anywhere in it.