Should you build a feature just because your competitor shipped it?

A button that would take your answers, send them somewhere, and hand back a paragraph telling you whether to build, kill, or prove the thing you were asking about. Then I looked at what I had built and realised it was the exact object Appray exists to replace: an authority handing you an answer, instead of eight rounds of questions that make you find the answer yourself. I deleted the branch and went for a walk, mostly annoyed at how far I'd got before noticing.

What actually happened

None of those three competitors added AI scoring because their users were begging for it in support tickets, as far as I can tell from the outside. They added it because the other two were adding it, and I know that because I do the same thing to them. I check their release notes more often than I'd admit to a stranger, and for about a day this summer that habit nearly talked me into contradicting my own product, inside my own product.

That's the part worth naming honestly. The pressure to match a competitor's feature rarely arrives as a business decision. It arrives as a low, specific anxiety that you're the one falling behind, and anxiety is a bad architect.

The reviews told me something I hadn't bothered to check

Before building anything I went and did the thing I tell other people to do first: I searched. Appray has a few hundred App Store reviews at this point, and I read through them for the word AI. Two mentions. One person asking, politely, whether the app secretly called out to a model despite what the listing said. One person saying they were glad it didn't, because they'd deleted three other apps in the category for exactly that reason.

That's not zero demand rounding down to zero. It's negative demand for the thing I nearly built, from the two people who bothered to mention it at all. The pressure I felt wasn't coming from my users. It was coming from three browser tabs I check too often, and I'd nearly let three browser tabs override a hundred reviews.

Hygiene, and a bet

Not every "the competitor has it" feature deserves the same answer, and pretending otherwise is its own mistake. Some things become table stakes fast. Export, a decent search, an undo button. If everyone in your category has them and you don't, you're not being disciplined, you're being behind, and users will read the absence as neglect rather than restraint.

The AI button wasn't that. It was a bet about how the job should be done, and the other apps and I disagree about the answer, which is fine, that's what competition is supposed to look like. The test I use now to tell the two apart is this: does not having it undercut the actual reason someone picked me over the bigger name, or does having it just make me look a bit more like the bigger name. Export failing that test would be a real gap. An AI verdict button failing that test just means I'd have spent a week building the reason to leave.

A second test, for when the first one is close: would a user who has never opened a competing app still describe this as something missing from mine. If the want only shows up after reading somebody else's changelog, it isn't a job. It's fear wearing a feature request's shape, and it deserves to be treated as fear, not as backlog.

The time I got this backwards

I'd like to end there, disciplined and vindicated, but the same instinct has cost me before, and I don't think the story is honest without it.

A competitor shipped a proper iPad layout in the spring. I filed the request under the same heading as the AI button: they're not solving a job I'm not already solving, they're chasing each other, ignore it. I sat on that for two months. Then I actually opened App Store Connect and looked at my own numbers instead of guessing, and eleven percent of Appray's downloads were happening on iPad, onto an interface that rendered at phone width in a small grey rectangle in the middle of a much larger screen.

That wasn't competitor parity. That was a real, sized gap in my own product that happened to arrive wearing a competitor's name on the request, and I nearly missed it because I'd already decided that anything shaped like "they have this too" was noise. The lesson isn't "ignore competitors" any more than it's "copy competitors". It's that the shape of a request tells you nothing about whether it's worth acting on. The size of the group behind it does, and you only see that by checking your own numbers, not theirs.

What I do differently now

I gave myself one rule after the AI weekend: I'm allowed to look at a competitor's changelog once a month, not on the day it publishes. It sounds small. It has already stopped me reacting to two more releases since, on schedule rather than on impulse, and both times the answer on the monthly check was calmer than it would have been on day one.

The roadmap is supposed to be a record of the jobs your own users are stuck on. The moment it starts tracking someone else's engineering calendar instead, you've quietly handed your product's direction to a competitor who has never seen your reviews and doesn't know your eleven percent.