18 Jan 2026 · Niran Kittisak

Cohort retention without a vanity D1

Week four of Cohort Telemetry Studio has a simple rule: you may calculate D1. You may not put it on the slide you would send upstream.

D1 is not evil. It is fast, cheap, and almost always available. That is why it becomes a trophy. A tutorial rewrite that inflates same-day return and does nothing by D3 will still get a congratulatory Slack message if D1 is the headline.

The homework

Students take two install weeks: one before a tutorial change, one after. They build retention through D7 (D14 if the title has the volume). Cuts required: install week and store. Optional: device class. Forbidden: a single blended curve used as “the number.”

Then they write three sentences a designer can act on. If the third sentence is “D1 went up,” they rewrite it. Alumni still complain that the SQL primers for this week are thin. They are right; we would rather you struggle on joins than on storytelling.

What “without vanity” actually means

It does not mean you ignore early signals. A collapse on D0 tutorial complete is still an emergency. It means you do not let an early bounce substitute for a cohort you have not waited on. Gaming App Analytics, as we practice it, is mostly about waiting with a labelled table instead of decorating impatience.

A mild outcome, stated plainly

In the idle-studio case on our Reviews page, post-rewrite install weeks sat about a point higher at D7. Nobody framed that as a transformation. They framed it as “the dictionary finally matched the tutorial.” That is the standard I want on a publisher call.

Full module list: Cohort Telemetry Studio.