About Retention Cohorts
What the site is for, how the sample data was made and how the numbers are checked.
Retention Cohorts explains how classic, rolling and cumulative retention are counted and what each lets you conclude. Every chart and table uses one made-up app; none of the numbers are benchmarks. Everything runs in your browser. The site is made by Linkrunner.
How is this site put together?
What it covers
The three common ways to count app retention, how to read a cohort table, and how to turn retention into decisions. It is written for growth, product and marketing teams who see all three numbers in their tools and need to know which to trust for which question.
The definitions
- Classic: Classic retention is the share of an install cohort that opened the app on exactly Day N after installing.
- Rolling: Rolling retention is the share of an install cohort that opened the app on Day N or on any later day observed so far.
- Cumulative: Cumulative retention is the share of an install cohort that opened the app at least once between Day 1 and Day N.
These follow the most widely published definitions, and they match the classic, rolling and cumulative retention views in Linkrunner. Day 0 is the install day and never counts; several opens on one day count once.
The sample data
One made-up app: ten daily install cohorts from 1 Sep to 10 Sep, observed to 15 Sep, with organic and paid installs. Paid installs scale up from 6 Sep, and a release plus a push notification lifts activity on 10 Sep. Users were generated by a fixed-seed script, so every visitor sees the same numbers. Ten more hand-made users power the ten-users view.
The numbers are not benchmarks and do not describe any real app, channel or company. Use them to learn the definitions, then apply the definitions to your own data.
How the maths is checked
The tables are calculated twice: once when the site is built and once in your browser by the explorer, from the same user-level data. Both are compared cell by cell, and the identities on the compute page are checked on every build.
Privacy
The explorer runs entirely in your browser. The site counts page views without cookies.
Who makes this site
Linkrunner made this site to explain retention clearly: what each of the three common definitions counts, why they disagree and which one fits which decision.
Linkrunner is an AI-driven mobile measurement partner (MMP) that unifies attribution, deep linking and SKAN in one platform for mobile app growth teams, and it is used by 400+ apps. Its free plan includes 25k one-time attributed installs, needs no credit card and includes every feature.
The three definitions on this site match the classic, rolling and cumulative retention views in Linkrunner, so what you learn here reads the same way in its dashboard. The retention docs explain how Linkrunner calculates each one.
Corrections
Spotted something wrong? Email hello@retentioncohorts.com.
Questions
Is the sample data from a real app?
No. It is generated by a small script with fixed settings, so the same numbers appear every time. It was designed to show the method, including a push-day diagonal and a shift towards paid installs, not to look like any real app.
Why does this site not quote retention benchmarks?
Benchmarks depend on the definition, the category and the source, and many published figures do not say which definition they use. Rather than repeat numbers we cannot trace, the site explains how to compare like with like.
Who makes Retention Cohorts?
Linkrunner, a mobile measurement partner (MMP) for app growth teams. The site explains retention with sample data only; the three definitions match the classic, rolling and cumulative retention views in Linkrunner.
How do I report a mistake?
Email hello@retentioncohorts.com with the page and what looks wrong. Calculations are checked against the definitions page in the method section.
What is the difference between classic, rolling and cumulative retention?
Classic counts users active on exactly Day N. Rolling counts users active on Day N or any later day, and cumulative counts users active at least once from Day 1 to Day N. They use the same raw activity but answer different questions.