Release Signal Lab
Event contracts, version-scoped funnels, and a live gate review on a build you actually intend to ship.
Course outlineBangkok · App release analytics
We sit with your event stream before the store button.
Cache Routehub is a small studio that trains product and QA pairs to catch broken funnels, silent crashes, and locale-skewed metrics while the build is still on a test track — not after a one-star week in Thailand.
Open the Release Signal Lab
127release gates reviewed in 2025
18product teams in Bangkok & Chiang Mai
11 daysmedian cycle from freeze to gate sign-off
2.3×fewer post-ship metric disputes after the lab
A recent gate
“Checkout started_at fired twice on Android 14 after a WebView bump. The funnel still ‘looked healthy’ in Mixpanel until we split by OS build.”
A Bangkok wallet team brought that trace into Release Signal Lab week three. They held the store submission, patched the double-fire, and republished with a version-scoped dashboard. The story is ordinary on purpose: most Analytics QA for App Releases is unglamorous instrumentation hygiene.
Read longer case notesProgrammes
Courses built around a real freeze date
Each programme assumes you already have an analytics vendor. We teach how to interrogate it before a release, not how to pick a logo.
Cohort Funnel Diagnostics
Find where new users vanish across app versions, then write a readout a PM can defend in a ship meeting.
See all programmes
Crash & Retention Audit
Separate crash-free session vanity from sessions a person could actually finish a task in.
See all programmesWhat we actually check
Analytics QA is not a prettier dashboard
If a metric cannot survive an OS upgrade, a locale switch, or a late schema change, it does not belong on a release gate.
Event contracts before code freeze
Names, properties, null rules, and who owns the diff when engineering renames a screen.
Funnels split by app version
A blended funnel that looks “stable” often hides a regression that only lives on the candidate build.
Locale, currency, and Thai store quirks
Baht formatting, Buddhist calendar dates, and dual-language onboarding all change event payloads.
Store listing experiments
Screenshot and subtitle tests that leak into in-app acquisition events get a dedicated QA pass.
From the lab floor
What teams say after a gate
The Event contract before code freeze module forced our Android and iOS leads to agree on checkout_started properties in writing. We still argue, but we argue against a sheet instead of a Slack thread that vanished.
Crash & Retention Audit made us stop quoting crash-free percent in the ship email. Slightly less flattering. Much harder to ignore.
Week four of Release Signal Lab assumed more SQL than our growth intern had. The schema still came out cleaner; just budget an extra coaching hour if the team is product-heavy.
Next Bangkok cohort
Bring a freeze date, not a hypothetical app.
Enrolment is informational on this site. Write to the studio and we will place you on a cycle that matches your store window.