Retention metrics in a predefined dashboard can differ when they use different customer populations, filters, time frames, or calculation logic. Compare the metric definitions and filters to determine whether the metrics are measuring the same thing.
Understand why metrics differ
The Retention Dashboard includes metrics such as repeat rate, last month's retention, 3-month cohort repeat rate, and average time between purchases. These metrics can use different filters, date ranges, and calculation types so that similar-looking values can produce different results.
A metric name can also suggest a customer retention percentage while the calculation measures revenue. For example, Last 6m retention rate % can use this calculation:
Last 6 months revenue from returning customers Ă· total last 6 months revenue Ă— 100
This result represents the share of revenue from returning customers. It doesn't represent the percentage of customers who returned.
For example, the 6-month retention metric can include an extra customer filter with an app URL condition, while the 12-month metric doesn't use the same filter in its customer filter aggregate. The metrics therefore can use different populations, which makes their values appear inconsistent. Aligning the filter logic makes the values consistent.
Compare the metric definitions
When you compare retention metrics, check the following settings:
Customer filters: Confirm that both metrics target the same audience.
Event filters: Confirm that both metrics use the same conditions, such as successful purchases or a specific source or URL.
Time frames: Compare periods such as 6 months and 12 months, and confirm that both metrics apply the same logic to each period.
Calculation type: Open each metric definition and check whether it measures customers, revenue, or another value.
Detailed definitions: Review the Evaluation section for more information, where available.
Align the metric logic
To make retention metrics comparable, align the customer filters, event filters, time frames, and calculation logic across the metrics you’re comparing. If one metric includes an extra customer or event condition, the results reflect different populations and aren't directly comparable.
Review the metric setup before interpreting the dashboard. When the definitions use the same population and logic, the values should behave as expected.