Why your retention curve lies in the first seven days

Product teams in the United Kingdom still walk into Monday with a D1 number as if it were weather. It is not weather. It is a mixture of store ranking, a push you sent because the chart looked lonely, and the simple fact that new users open an app to see whether it installed.

In Retention Cartography we ask students to hide days 0–7 and describe the rest. The remaining curve is usually flatter, occasionally more honest, and almost always politically inconvenient. A grocery loyalty app we sat with last autumn had a celebrated D1 of people retrying a failed barcode scan. The “retained” user was stuck.

App Analytics tools make the opening week beautiful because the sample is large and the story is simple: we acquired, they came back. From day eight you need a dormancy definition. You need to know whether “open” means a foreground session or a silent refresh. You need to stop averaging users who completed onboarding with users who bounced from a permission dialog.

None of this requires a new dashboard. It requires a sentence under the chart: “Week one includes store-driven reopen; we read habit from day eight.” If your leadership refuses the sentence, the curve is not lying to them. It is being used.

If you want the table exercise we use in the room, it lives in the Retention Cartography notes. Bring a cohort that embarrassed you.