Thailand · 18 January 2026
Thailand seasonality is not a footnote on a Day-14
Playbooks written for US or European calendars arrive in Bangkok looking confident. They treat Songkran as a footnote, year-end commerce as “Q4,” and late-night usage as a rounding error. App Retention Cohort Analytics built that way will invent product failures that were actually people leaving the city.
We teach students to annotate, not to delete. Deleting a holiday week makes the chart prettier and the comparison dishonest. Annotating it lets a CPO see the dip and still trust the window.
Songkran is a mobility event
For many consumer apps, Songkran is not a feature problem. It is a change in where people are and what they open. A Day-7 built on first-open during that week will move. A Day-14 that compares Songkran acquisition cohorts against a quiet February cohort is not a regression analysis; it is a calendar mismatch.
In a Window Clinic last year, a marketplace team paused a checkout experiment because Day-7 “collapsed.” Reconstructing the stamp to first successful order, and marking the holiday week on the table, changed the meeting. The experiment resumed. The lasting artefact was a one-line rule: holiday windows are labelled in the readout.
Year-end commerce is not universal
E-commerce peaks are loud in December. Travel and education products may peak elsewhere. One atelier student in a travel-adjacent app noted, fairly, that our house examples lean on Songkran and year-end, while their category moved in July. That reservation belongs in the record. The method still holds: name your local peaks, compare like-for-like calendars, do not import someone else’s quarter.
Late night is a usage shape
Bangkok usage often stretches later than the dashboards’ default day-boundary. If your “day” cuts at midnight UTC, you will mis-bucket evening sessions. Clock-start is not only an event name; it is also a timezone decision. Write it down.
The Bangkok Market Retention Briefing is a single afternoon on these distortions. The atelier spends a full evening on them. If your model was trained on another country’s calendar, start there before you rewrite the product story.