Daily active users rewards opening. Opening is cheap. A widget, a badge, a weather tile, a streak panic — all of them can lift DAU while the verb that justifies the charge stays flat. We have sat in rooms where DAU rose through a month in which trial-to-paid fell. The slide still went green because green was the colour assigned to “up.”
The week-0 costume
Install week is a carnival. Push permission prompts, a tour, a discount, perhaps a widget already on the home screen. Counting those people as the same kind of “active” as a subscriber in month four is how averages become costume jewellery. In Retention Signal Architecture we rebuild one curve after removing week 0 and after requiring first value. The line gets smaller. The conversation gets adult.
Use DAU as a smoke alarm, not a north
If DAU collapses on a Tuesday, something in delivery probably broke. That is a useful alarm. It is not a strategy. For subscriptions we would rather see: first value inside the agreed clock; returned_in_window; paid_or_renewed with refunds sitting beside it. Those verbs are slower to plot and harder to game with a badge.
Founders dislike smaller numbers. We know. The Glasgow language-app case on our reviews page retired a streak metric for the same reason. The all-hands slide looked worse for a fortnight. Trial messaging then pointed at a spoken lesson instead of a mascot. Nobody called it a miracle. They called it less embarrassing.
If your app is genuinely a daily utility — a commute card, a prayer time, a glucose log — DAU may earn its place. Say so, and define the verb that makes a day “active,” or you are counting wallpaper again. The Atlas has a line for that argument. Bring it to class.