What is a retention curve?
Used in a sentence: The new onboarding didn't change the early drop much, but the retention curve flattens higher now, which is what we wanted.
How it works
Start with a cohort, usually everyone who installed in the same week. For each day or week after install, count how many of them were active and divide by the cohort’s starting size. Plot those values in order and you have the curve.
The steep early drop shows users who tried the app and did not find value. The later flat section shows the users who stayed. Teams compare curves across cohorts to see whether product or onboarding changes moved the early drop, the flattening point, or both.
Example
A language learning app plots weekly retention curves for cohorts before and after adding a daily lesson streak. The early drop looks similar for both. The newer cohorts’ curve, however, flattens at a visibly higher level and stays there. The team concludes the streak did not help first impressions but did help users who got past the first week build a habit.
Retention curve vs churn rate
| Retention curve | Churn rate | |
|---|---|---|
| Shows | Share of a cohort still active over time | Share of users lost in one period |
| Format | A line across many intervals | A single number |
| Best for | Spotting where and whether users settle | Tracking loss month to month |
Churn rate is a snapshot. The curve shows the full shape of how users leave and stay.
Best practices
- Use a clear definition of active, tied to the core action, not just an app open.
- Compare cohorts of similar size and season.
- Focus on where the curve flattens, not only on D1.
- Annotate product and campaign changes so shifts can be traced to a cause.
How to do this with Digia Engage
To lift the early part of the curve, growth teams target new cohorts from their CEP with onboarding walkthroughs, first-action nudges and checklists built in the no-code builder. Habit features such as streaks help the later part. Each experience goes live without an app release, and teams compare the resulting cohort curves in their analytics. See the Activation use case.
Frequently asked questions
- What does a good retention curve look like?
- A good curve drops early, as some users always try an app and leave, then flattens into a stable line instead of sliding towards zero. The flatter and higher that line, the stronger the retained core. What counts as good varies widely by category, so compare against your own past cohorts first.
- What is D1, D7 and D30 retention?
- They are the share of a cohort that returns to the app on day one, day seven and day thirty after install. D1 reflects first impressions and onboarding, D7 shows whether early value landed, and D30 indicates whether a habit is forming. Together they summarise the shape of the curve.
Related terms
Read more
Part of the In-app engagement glossary.