The 30-day window isn't magic — it's just enough time to harden the rough edges without the project losing momentum.
Week-by-week
Week 1 — Define ready
Pick the metrics you'll use to call it "ready" and write them down. If you can't measure it, you can't decide it's done.
Week 2 — Harden
Add error handling, timeouts on every network call, and a thin layer of telemetry. Don't chase coverage — chase confidence.
Week 3 — Soak
Run realistic traffic against it for a few days. Watch the metrics from week 1, not your gut.
Week 4 — Ramp
Roll out behind a flag. Ramp gradually — 1% → 5% → 25% — and watch the same metrics at each step.
What gets dropped
Plenty. The goal is shipping something that works for real users, not shipping the platonic ideal of the feature. The ideal version comes later, after you have real usage data.
"Production" isn't a code-quality milestone. It's a decision to expose your work to the kind of feedback you can't get any other way.
