Thailand · 28 July 2026
Reading Thai app cohorts after Songkran
Holiday silence and the return spike impersonate a messaging win. If the calendar is not on the chart, your lifecycle story is weather, not craft.
Every April we see the same review. Sends drop during Songkran. Opens look strange. The week after, sessions return. Someone in CRM claims the “welcome back” journey worked. Someone in product says the app healed itself. Both can be wrong at once.
The holiday is not a weekend
Songkran is a multi-day interruption with travel, water, and family obligations. Device mute rates rise. Delivery still succeeds. Attention does not. If your quiet-hour policy only knows clock time, it will keep treating 14:00 on a travel day as a normal slot.
Payday in Thailand is another recurring weather system. Mid-month dips in wallets, ride-hailing, and grocery look like churn if you do not draw the pay cycle on the cohort. Cohort Threading spends a full session on this annotation. It is not folklore; it is a join to a calendar table you maintain yourself.
What we ask students to do
Before any post-holiday recap, mark: public holidays, regional observances that affect your user base, payday windows, and major marketplace sale days. Then compare messaging-touched users to a holdout across those same marks. If the holdout also snaps back after Songkran, your journey did not cause the snap.
A Bangkok payments pod in Signal Path Studio did this for an “inactive wallet” path. The SMS step after Songkran was present for the return and irrelevant to it. They kept the in-app card and dropped the SMS. That decision only looked obvious after the calendar was visible.
US-centric holiday tables are not a shortcut
Imported analytics templates often ship with Thanksgiving and Black Friday. They do not ship with Songkran. If your warehouse uses an imported holiday dimension, audit it. Devpacketbase sits in Yannawa because this is not an edge case for us. It is the default year.