What it is.
Five steps. Define the retention behavior, which is the event that means somebody came back. Choose your measurement, add filters, add breakdowns, then pick the visualization.
Custom retention brackets are the step worth learning. You can ask how many users come back between days 15 and 30, which tells you about habit where a standard curve only shows the drop.
Breakdowns are what make it actionable, because retention by plan, by acquisition channel, or by first feature used usually differs sharply.
What you get.
- A retention definition you set, based on the event that means returning.
- Custom brackets, so you can measure one specific window instead of a whole curve.
- Filters to narrow the population you are measuring.
- Breakdowns by plan, channel, or property.
- Measurement options including revenue and property averages per user.
How to set it up.
Pick the event that genuinely means somebody came back, since a login is weaker than a core action.
Set the measurement, choosing unique users unless you have a reason to count something else.
Filter to the population you care about, because retention across all signups hides the segments.
Add a custom bracket for the window you care about, such as days 15 to 30.
Break down by acquisition channel and by plan, then compare the curves.
Save the report and re-read it monthly, since retention changes slowly and needs the same definition each time.
Use cases
Define what returning means
Pick the event that counts and build the report on it.
Brackets that match your cycle
Set custom windows instead of accepting the default curve.
See past the flattened line
Split the report so a weekly product is not measured monthly.
Best for
Comparing acquisition channels
Two channels can deliver the same volume and completely different retention.
Measuring habit, not curiosity
A custom bracket at days 15 to 30 answers a different question than day 1 retention.
Onboarding experiments
Retention by first feature used tells you which onboarding path to push people down.