I built the feature people asked for most. Then came four months of silence.

Appray doesn't track anything. No accounts, no analytics SDK, nothing phones home, ever. I still think that's the right call and I've said as much elsewhere on this blog. What I hadn't properly worked through until this year is what it costs on the other side of shipping. You lose the easy, lazy way of finding out whether the thing you built actually did anything. Most apps can open a dashboard and see a number go up or stay flat. I can't. I built the one kind of product where that shortcut doesn't exist, and I didn't think about what would happen the first time I badly wanted it.

The request itself was simple and kept arriving in different words. People finish the eight round interview, get their verdict, Build or Kill or Prove, and then a chunk of them want to take it somewhere else. Put it in a deck for a co-founder. Paste it into a notes app. Print it and pin it above a desk, which one person told me they'd done and which is a far lovelier image than I expected from what is, underneath, a decision tool. Thirty four separate emails over about five months asked for some version of getting the verdict out of the app. That's a real, countable number, and it's the second highest I've had for any single request behind wanting an Android version.

What I built

Export as a PDF. One button on the verdict screen, a clean layout with your answers on one side and the verdict and reasoning on the other. It took about three weeks including two rounds of fighting the page breaks into behaving, which is its own small, boring horror story I won't get into here. I shipped it in a March update, mentioned it in the release notes, nothing more than that. No launch post, no email to the thirty four people who'd asked. In hindsight that last part was a mistake, and not just for the obvious reason.

Then, nothing

No complaints. No bug reports. No more emails asking for it, which I first read as a good sign, the kind of quiet you get when something just works and nobody needs to write in about it. Four months went past before it properly occurred to me that silence and success look identical when you have no way to tell them apart, and I had built exactly the kind of app where you can't tell them apart. A normal app owner would glance at a usage graph and know in ten seconds. I had a button that might have been tapped forty times or might have been tapped never, and both of those were consistent with everything I'd observed, which was nothing.

Asking by hand

So I did the only thing left to me, which was ask people directly. I went back through the original thirty four emails and picked twelve where I still had a working address, all people who had specifically asked for export rather than just anyone on the list. Short message, nothing salesy: you asked for this a while back, I built it, did it end up being useful, I genuinely want to know either answer.

Seven replied. Two had used it more than once and were glad it existed. One sent a screenshot of the exported PDF sitting in a shared drive folder next to actual investor materials, which pleased me more than it probably should have. Three had used it exactly once, right after it shipped, and never again, not because it failed but because once the decision was made they no longer needed a document about it. The deciding was the whole job. The PDF was just something to hold onto during it. And two, this is the part that has stuck with me, had completely forgotten they'd asked for it and didn't remember the feature existed at all, despite having opened the app more than once since March.

Request and use are not the same claim

That third group is the one I keep turning over, because I recognise the shape of it from a different angle now. I've written before about the gap between how many people ask for something and how many actually need it. What I hadn't clocked properly is that request and use differ in timing as much as in headcount. Someone emailing me about export was, more often than not, asking in the exact moment they wanted proof that the whole exercise had been serious, that they hadn't just spent twenty minutes poking at an app. That feeling is real and worth designing for. It just doesn't reliably survive as long as the feature itself does. By the time export was sitting there ready to use, the moment that had made someone want it had often already passed, and the request they'd sent me weeks earlier was the only trace left of a need that had come and gone on its own.

I'm not killing the feature. It costs almost nothing to keep, it does its job for the people who do reach for it, and two glad replies out of seven on something that took three weeks is not a bad return by any measure I'd normally use. But I've stopped counting those thirty four emails as evidence the feature is working. They were evidence the feature was worth building. Those are two different claims, and for four months I'd been quietly treating them as the same one.

What I'm actually changing

Nothing as grand as a new process, more of a habit. For anything I ship because enough people asked for it, I now put a date in my calendar roughly ten weeks out to go back and check with a handful of the people who asked. Not a survey, not an in-app prompt I don't believe in anyway, an actual email like the one I sent in July. It's slow and it doesn't scale past an app this size and I don't especially care, because the alternative I'd been running on, silence equals success, was never a measurement at all. It was the absence of one, and I'd been reading it as an answer only because it was the only signal I had.