Method
Analytics QA for App Releases
A release is measurable or it is a rumour. This page is the studio’s working definition of Analytics QA — the same ritual we teach in class and use on external gate reviews in Bangkok.
What “QA” means here
Functional QA asks whether a button works. Analytics QA asks whether the story you will tell about that button after launch can be defended. If checkout_started cannot be trusted on the candidate, you do not have a conversion narrative. You have a hope.
We do not replace your test engineers. We sit beside them with a contract sheet, a version filter, and a bias toward hold when the evidence is thin.
The gate in one hour
- Contract: which journey is in scope this freeze.
- Evidence: version-scoped funnel, crash vs usable-session, locale slice.
- Known gaps: listed, owned, dated.
- Decision: ship, hold, or hold-for-sample.
Guests from PM or eng are welcome. Slides are optional. The memo is not.
Thailand-specific checks
Locale is not a polish pass
Shipping into Thai stores without slicing th-TH is how northern users disappear from a warehouse join.
Currency shape
Baht as satang integers, decimal baht, or symbol-prefixed strings — pick one per property and diff both OS wrappers.
Calendar fields
Buddhist era years mixed with Gregorian ISO will look like “bad data quality” when they are actually two clocks.
Bilingual onboarding
A step that exists only in one language still needs an event, or your funnel will invent drop-off.
Store metadata leakage
Subtitle and screenshot experiments must not masquerade as in-app product wins.
Learn it, or hire the ritual
Most teams take Release Signal Lab once and then run gates themselves. Watch Floor exists when freezes are frequent and you want a quarterly schema diff without rebuilding the teaching cycle.
If you only need the method on paper, start with the pre-launch checklist and the hold-for-sample note. If you have a freeze date this month, write to the studio.
Browse programmes