Three cohort composition traps we see in Bangkok product teams

Composition is the quiet villain of cohort work. Teams in Bangkok often ship a feature, then open a monthly retention chart that moved for reasons unrelated to the feature.

Trap one: weekday versus weekend installers. Weekend cohorts may contain more casual browsers. If your release landed on a Friday, the next week’s curve is not a clean A/B.

Trap two: geography mix. A promotion in a new city can pull users with different payment methods or delivery expectations. Aggregate Thailand retention then hides a city-level drop.

Trap three: returning users miscounted as new. Identity stitching after a login redesign can inflate “new” cohorts with people who already knew the product. Their high return rate flatters the curve and misleads the roadmap.

Before attributing a retention change to product work, ask: did the people inside the cohort change? If yes, re-cut the table before you brief leadership.