Writing/App Store
The last two weeks before review
Everything I thought was left was small. None of it was the work. The work was the compliance surface nobody demos and every reviewer opens first.
By the beginning of March, TeamTipOut did everything it needed to do. The split was right, the crews worked, the export looked good. I thought I had about a week left. What was actually left was two weeks of work that appears on no roadmap and in no demo, and that a reviewer finds within four minutes.
The blockers a reviewer finds first
On iPad, taps were passing straight through a glass layer to whatever sat underneath it. Modals were sizing themselves for a phone. A name prompt could be skipped in a way that left the account half configured. All three were on the largest device the app supports, which is the one a reviewer opens first and the one a phone-shaped app gets tested on last.
Reviewers do not use apps the way users do. They go around the edges: rotate it, background it, skip the thing you cannot skip, open it on the largest device you support. Rejections come from that perimeter, not from the feature you spent the month building.
The compliance surface
A web page for account deletion, because a store requires one to exist outside the app. Terms updated for subscription rules. A privacy declaration that actually matches what the app collects rather than what I assumed it collected. Deep links registered on both platforms. None of it is a feature and all of it is between you and being live.
“None of this is on a roadmap, and all of it is between you and being live.”
The things you only get one chance at
Source map uploads had to be configured before the first production build, not after, because a crash report from a release you cannot symbolicate is a stack of hex. Rate limiting on the one-time code had to exist before anyone could find the endpoint. A cache leak had to be fixed before real sessions started stacking up in it.
That is the category worth planning for: work that is cheap before launch and expensive or impossible afterwards. It is a short list and it is never the list you are excited about.
What I would do differently
I would run the review perimeter at the halfway point rather than at the end. Borrow the biggest device, rotate everything, background the app mid-flow, try to skip every required step, and read the store's own checklist as if it were a bug report someone had already filed. Two hours in the middle of the project, instead of two weeks at the end of it.
