Storefront layer
Rank, conversion, screenshot tests, and localisation of the listing itself. A Thai-language variant is a product decision, not a translation ticket.
Field note
Analytics for regional app expansion, written from the market you are entering.
This page is the studio’s working lens. It sits underneath every course. If you only read one thing before the syllabus, read this.
A region is not a filter in your BI tool.
Teams arriving in Thailand often inherit a dashboard with a “SEA” toggle. That toggle hides the thing that matters: storefront policy, language variants, prepaid top-up rhythm, and which payment rail actually completes. Expansion analytics starts by naming those differences before any spend model.
We treat Thailand as a home market with its own iOS/Android mix, PromptPay expectations, and festival calendar. Vietnam, Indonesia, and the Philippines enter as comparison corridors, not siblings in a blended average.
What we actually measure
Four layers that refuse to collapse.
Acquisition layer
Windows, privacy constraints, and channel mix that still make sense when SKAN returns sparse. Last-click stories get retired early.
Retention layer
Cohorts aligned to local paydays and festivals. A 28-day window that ignores Songkran or payday is a decorative chart.
Monetisation layer
Wallet, card, and carrier billing sit beside ARPU. If checkout fails on a popular rail, the funnel is not “soft demand”.
Common failure modes
Patterns we see before a launch goes quiet.
Imported North American event names
Global taxonomies that never added a PromptPay success event, or that treat “subscribe” as if every market has the same billing cadence. The corridor track rebuilds the taxonomy with local rails first.
Top-chart theatre
A spike on a Thai chart that coincides with a featured placement or a category reshuffle. We teach a simple test: if organic search intent did not move, do not brief the spike as product-market fit.
One creative, four captions
Creative testing that only swaps headlines while screenshots, currency, and social proof stay Singapore-coded. Localisation telemetry is part of analytics, not a brand side quest.
Exec decks without limits
A board slide that cannot name what the data cannot support. Our scorecards require that sentence. It is the difference between a teaching studio and a cheerleading vendor.
If this lens matches the problem you actually have, start with the corridor track.
Or write to the studio with the market, the date, and the stack you already use.