Journal · 11 March 2026
DAU is not a product metric
Daily active users counts doors opening. Subscription apps, in particular, need an account of rooms occupied — and of rooms quietly abandoned after billing.
App analytics inherited DAU from advertising and social products, where showing up was the business. A paid subscription is a different contract. A customer can open your application every morning to dismiss a badge and still be one notice away from churn. Calling that person “active” is a courtesy to the chart, not to the P&L.
The usual defence is that DAU is “top of funnel for engagement.” That sentence does work only if you have already defined the behaviour that constitutes use. For a meal-kit, use might be an order placed. For a regulated wallet, it might be a completed payment, not a balance check caused by anxiety. Until that definition exists, DAU is a headcount of curiosity and push notifications.
In the Retention Ledger we ask students to pick an entry event that is not a session start. The first draft is often still a login. Logins are cheap; they follow password managers and biometric convenience. The second draft is usually better: a completed job, a saved recipe, a submitted claim. That event becomes the row heading of the ledger. DAU can remain on an engineering dashboard as a load signal. It should not appear in a board pack as evidence of product health.
There is a political cost. Stakeholders like a number that moves every day. A four-week cohort ledger moves slowly and sometimes downwards. That is the point. If your organisation cannot tolerate a slow true number, it will keep DAU and call the resulting surprises “seasonality.”
We are not arguing for monastic products that nobody opens. We are arguing that opening is not the work. Measure the work. Then, if you must, glance at DAU to see whether the push system is shouting.