Songkran cohorts and false churn
Every April a commerce app somewhere in Thailand writes a board memo about lost brand love. The curve looks like a cliff. The calendar looks like water fights, travel, and closed checkouts. Cohort Retention Mapping exists so those two facts stay in the same sentence.
A retention chart without labels is a Rorschach test. If your April cohort bought heavily in the week before Songkran and then went quiet, that is not automatically churn. Many shoppers are on the road, on prepaid data, or standing in someone else’s kitchen. 11.11 and payday Friday create the opposite illusion: a spike that the following Monday cannot repeat.
The studio method is almost rude in its simplicity. Draw the cohort. Overlay public holidays and known payday windows. Compare the same cohort’s next non-holiday week before you write the memoir. If the curve recovers, you had weather. If it does not, then you may have a product or assortment problem — still not a licence to buy more traffic until the path can hold it.
Classroom datasets at Switch Nodehub are mostly commerce, not subscription video. That is another limitation we state on the catalogue. A streaming app with a monthly bill has different Songkran physics. We will not pretend the commerce worksheets transfer cleanly.
Verified learners on Lattice Studio have called this module the reason they stopped apologising for April. That is enough. If your next board pack still treats the holiday as a character flaw in the customer, bring the chart to the studio before you rewrite the roadmap.