Journal · 9 September 2025
The n-day retention trap in mobile
Mobile analytics inherited D1 / D7 / D30 from games that wanted a daily habit. Plenty of useful apps are not daily habits. A travel organiser, a festival wallet, a parental-leave checklist — people return in bursts. Measuring them on D7 is a category error dressed as industry standard.
The trap is worse when “returned” means any session. A crash-restart can look like love. In class we force a retained event that matches the job: a saved itinerary, a ticket scan, a completed checklist item. The n-day number usually falls. That fall is information.
Tenure weeks still help mobile teams, even if the public blogosphere prefers days. A week absorbs timezone noise and overnight sessions. You can still compute D7 as a diagnostic for onboarding; just stop promoting it as the quality of the product.
Store listing experiments will keep chasing D1 because app stores reward early return. That incentive is real. It is not the same as whether the product is worth keeping. Caption the chart with the incentive, or you will optimise for the store and call it craft.