A retention curve shows the share of a group of app users who return in each period after they first appear in your data. Plot the return rate on the vertical axis and time since the cohort began on the horizontal axis; the curve shows how quickly activity falls and where it levels off.
What a retention curve shows
Each curve represents a cohort: users grouped by a shared starting period. A weekly curve uses weeks since the cohort began, while a monthly curve uses months. Period 1 is the first period after that starting period.
The vertical value is the percentage of the cohort with activity during that period. A curve that drops quickly shows fewer cohort members returning in later periods. A flatter section means the share active from one period to the next has changed less. The line describes the cohort in your data; it is not a forecast of future activity.
Be precise about the cohort start. If your records establish install dates, you can group users by install period. If you only have purchases or other activity records, the first recorded row is the start you can measure. Calling that an install cohort would overstate what the data says.
How to calculate retention by cohort
ActivationPal’s cohort analysis tool turns a CSV of purchases or orders into a weekly or monthly return table. Add a customer identifier and a date for each row. The tool recognizes common customer and date column names; if it cannot identify them, choose the two columns in the controls.
For each customer, the earliest dated row sets their cohort. The table groups customers by that first week or month, then counts a customer as returning when they have at least one row in a later period. Each cell is the returning count divided by the number of customers in that cohort. The tool reports those percentages by cohort and period, rather than producing a line chart.
Choose weekly periods for frequent activity or a closer view of early returns. Choose monthly periods when returns are less frequent or you want to compare longer intervals. Keep the period consistent when comparing cohorts; changing from weeks to months changes the question the curve answers.
How to plot a retention curve
Use the table’s rates as the chart data:
- Choose one cohort row and copy its period labels and retention percentages into a spreadsheet.
- Set the horizontal axis to periods since the cohort began: period 1, period 2, and so on. Set the vertical axis to the corresponding retention percentages.
- Insert a line chart and format the vertical axis as a percentage. Leave periods with no result blank instead of entering zero.
To compare cohorts, add one series per cohort and align each series by age, so every series starts at its own period 1. Keep the cohort labels with the series. A calendar-week chart answers a different question because each line position then refers to a date rather than time since cohort start.
How to read a retention curve
Read the early decline first. A steep drop between the first few periods means a smaller share of the original cohort is active in those periods. Compare cohorts at the same age to see whether newer groups return at different rates than older ones.
Look for where the line begins to flatten. The percentage at that point is the share of that original cohort with activity in those periods; it does not mean the remaining users are continuously active. In the cohort tool, a customer can count again after a period without activity, so a later rise means more customers in the cohort had a recorded row in that period than in the preceding one.
Check how many customers are behind each line. A percentage from a small cohort can move more when a few customers return or stop returning. The tool shows cohort size alongside each row, so read that count with the curve rather than comparing percentages alone.
Recent cohorts may not have reached later periods yet. Leave those points out of the chart instead of treating them as zero retention. The cohort table leaves future periods blank and outlines a current period while it is still in progress.
Retention metrics to read alongside the curve
The curve gives a period-by-period view, while an average can hide how cohorts differ. ActivationPal’s table also reports average period 1 retention across cohorts with a complete first return period. Use that value as a compact check, then inspect the rows to see which cohorts sit above or below it.
Keep the denominator visible in your interpretation: a cohort’s retention rate is a share of the customers assigned to that cohort. Compare like periods, and keep acquisition or signup cohorts separate from cohorts defined by first purchase if those dates are available in your records. For more on the cohort view, read cohort analysis for app conversion and how to measure onboarding completion. You can also see how ActivationPal measures app conversion across onboarding, paywalls, and revenue.