Event contract before code freeze
Name grammar, required properties, null policy, and a signed owner when a screen is renamed. You produce a one-page contract for the journey you care about this cycle.
Flagship course
Release Signal Lab
Six weeks with a freeze date on the wall. You bring a candidate build; we teach the Analytics QA for App Releases ritual that decides whether that build is measurable enough to ship.
Who it is for
Pairs of product and QA, sometimes with a data engineer in the room. You already instrument events. You still discover, after a store push, that a property was null on Android 14 or that Thai locale users never entered the “success” step because the currency string changed shape.
It is not a beginner SQL bootcamp and not a vendor certification. If you have never logged an event, start elsewhere; if you have shipped three messy versions, this lab will feel uncomfortably specific.
Learning outcomes
- Write an event contract that engineering will actually diff against.
- Rebuild one core funnel scoped to app version, OS, and locale.
- Separate crash-free vanity from usable-session rates.
- QA a store listing experiment without contaminating in-app acquisition.
- Run a one-hour gate review that ends in ship, hold, or hold-with-patch.
Modules
Six weeks, one candidate build
Funnel integrity across app versions
Stop blending last quarter with the candidate. We teach version filters, late-arriving events, and how a WebView bump double-fires steps.
Crash-free versus usable-session QA
A session can be crash-free and still drop the user in a frozen pay sheet. You will pair crash vendors with task-completion events.
Store listing experiment readout
Icon and screenshot tests leak into first-open events. This week isolates listing traffic so product analytics does not “win” an experiment the store already decided.
Thailand locale and currency instrumentation
Baht symbols, Buddhist calendar dates, and bilingual onboarding change payload types. Built with Bangkok store windows in mind.
Gate review ritual
A live 60-minute review on your build: evidence, known gaps, and a written ship / hold decision. Guests from your PM or eng manager are welcome.
Instructor
Nalinee Sutham
Nalinee led instrumentation QA inside a SEA super-app before founding Cache Routehub’s teaching practice. She still sits on external gate reviews for Bangkok fintech teams when a release touches payments or KYC.
Her teaching bias is documentary: if it is not on the contract sheet, it did not happen. Students sometimes find that dry. The store does not grade charm.
Informational fee
Release Signal Lab is listed at ฿52,000 per seat for the Gate Retainer cycle, or included inside Watch Floor for teams running more than one freeze a year. Figures are published so you can budget. This website does not take payment.
Compare studio tiers on the pricing page. Refund rules live on refunds.
Request a cycleA limitation you should know
The lab will not instrument your app for you. If your SDK is missing, or if engineering cannot ship a debug build to TestFlight / internal track by week two, we pause the seat rather than pretend a slide deck is QA. That pause is not refunded after week one unless the studio cancelled — see refund terms.
Questions we hear before week one
Can one person attend without a QA partner?
Yes, but the gate review is weaker. The contract sheet needs someone who can change code. Solo PMs finish the exercises; they rarely close the freeze without an engineer in the last two sessions.
Do you support only Firebase?
No. Mixpanel, Amplitude, and warehouse-only setups (BigQuery / Snowflake via a thin event bus) are all in scope. We do not cover homegrown pixel-only web analytics; this lab is for native or hybrid apps.
Is there a certificate?
You receive a gate memo signed by the instructor for the build you reviewed. We do not sell framed certificates. Hiring managers who know this work ask for the memo and the contract sheet.
What if our sample size is too small to decide?
That is a valid gate outcome. Week six includes “hold for sample,” which is still a decision. Read When sample size says wait.
Voices from this course
Funnel integrity across app versions was the week that paid for itself. We had been celebrating a blended activation rate that quietly died on the candidate iOS build.
8 / 10
Release Signal Lab · “Strict about homework. Locale week was more useful than I expected. Would have liked a longer SQL primer before taxonomy diffs.”