How rolling retention works
Rolling retention, sometimes called return retention, is the percentage of an install cohort that opened the app on Day N or any day after. If you are measuring Day 7, any user who returned on Day 7, Day 20, or Day 90 counts. It is the most forgiving of the three methods because it treats a single late return as evidence the user is still around.
The formula follows the same pattern: divide the installs that count as retained on Day N or later by the cohort size, then multiply by 100. Rolling retention is always equal to or higher than classic retention, and it can be higher or lower than cumulative depending on your users, because the two count different groups.
Why rolling retention keeps changing
The defining property of rolling retention is that it is unbounded. Because it counts a return on Day N or any later day, a cohort's rolling Day N value keeps rising as users come back weeks or months later. This has two practical consequences that trip up growth teams.
First, older time periods almost always show higher rolling retention than recent ones, simply because older cohorts have had more time for late returns to accumulate. A recent period can look worse than an old one for this reason alone, not because retention actually dropped. Second, a rolling number you record today can go stale. If you export it into a report, it will keep climbing afterwards, so it will understate the final value unless you refresh it. Cumulative and classic retention do not have this problem, since both are bounded.
When to use rolling retention
Use rolling retention to gauge long-term stickiness, since it captures users who stayed loyal even if they skipped the exact measurement day. It is well suited to apps with irregular but durable usage, where the question is less "did they open the app on Day 7" and more "are they still a user at all." Just remember to refresh the figure regularly and to avoid comparing a fresh cohort against a mature one.
In Linkrunner, rolling retention is available for Day 1 and Day 7, and is returned as rolling_retention in the Campaign Reporting API.