Retention for weekly-use vs daily-use apps

What is usage cadence and why does it matter for retention?

Usage cadence is how often a typical engaged user naturally opens your app, and your retention metric should be measured at a cadence that matches it.

Usage cadence: the natural rhythm of use. A messaging or news app is daily; a grocery, banking or fantasy sports app may be weekly; a travel app may be monthly.

Classic Day N retention: the share of the cohort active on exactly Day N. It is the strictest of the three retention types.

A daily-use app and a weekly-use app can have equally loyal users and very different classic Day 7 numbers. The metric is not wrong; it is answering a question that does not suit the app.

Does classic Day 7 retention punish weekly-use apps?

Yes. A user who opens the app once a week is only counted on classic Day 7 if their usual day happens to fall exactly seven days after install, so loyal weekly users are missed by chance.

The site's ten-users example (a made-up app, sample data only) shows this. User 2 opens the app on Days 1, 3, 5, 8 and 12: clearly engaged, but not on Day 7.

Across all ten sample users, Day 7 reads 30% classic, 60% rolling and 80% cumulative. Same people, same activity, three very different answers. You can step through each user on the 10-users view.

Better measures for weekly-use apps

Cumulative windows

Cumulative Day 7 asks "did they come back at least once in the first week?", which is the natural question for a weekly app.

• It is bounded, so the number is fixed once Day 7 has passed.
• It never falls as N grows, so use it for goals like "come back within seven days".
• It hides how often people return, so pair it with a frequency measure.

Weekly cohorts with week-N retention

Group installs by week instead of day, and count a user as retained in week N if they were active on any day of that week.

• Week 1 retention for a weekly app behaves like Day 1 retention for a daily app.
• Weekly cohorts are larger, so each cell is less noisy.
• You lose day-level detail, which matters less when usage is weekly anyway.

Rolling, with care

Rolling Day 7 also counts user 2, but it keeps rising as late returns arrive and older cohorts have had more days to return. Use it to see eventual return, not to compare cohorts of different ages. The reading guide on flattening explains when a curve has settled.

Day-of-week effects

When usage clusters on certain days, install day changes what Day 7 lands on.

• A weekly app used mostly on weekends will show Day 7 that rises and falls with the install weekday.
• Classic Day 7 for a Saturday cohort lands on a Saturday; for a Tuesday cohort it lands on a Tuesday. Expect them to differ.
• Compare cohorts that started on the same weekday, or average across a full week of cohorts.
• Push notifications or releases on one calendar day lift one diagonal of the cohort table. See reading diagonals.

Matching the measure to the cadence

Stickiness is a useful complement. The DAU/MAU ratio shows how many monthly users show up on an average day. A weekly app will naturally sit lower than a daily app, so judge it against your own trend, not against daily apps.

Comparing yourself with published benchmarks

Benchmarks only mean something when the definitions match.

• Published benchmarks usually use classic, exact-day retention. Compare your classic numbers to them, not your cumulative or rolling numbers.
• Check the start event (install, not first purchase) and the return event (any open).
• Check the category. A weekly-use category will usually sit below a daily-use one on classic Day 7.

This explainer on what a good Day 7 retention rate looks like covers how to read Day 7 fairly, and you can browse category benchmark context for apps similar to yours. Whatever you compare against, match the definition first.