Study first
Review the ideas behind the questions
Review the retention measurement choices that make cohort and churn reports easier to explain. Focus on the return action, usage interval, cohort type, lifecycle stage, and what the metric can or cannot prove.
Choose The Retention Window
Retention windows should match how often customers are expected to come back and perform the value action.
- A product usage interval is the frequency with which you expect people to use the product.
- Retention numbers can be misread when the analysis does not match the product's natural usage interval.
- The critical event should define the return action used to count a user as retained.
In Practice
Daily Retention Can Mislead Some Products
A weekly ordering product may look weak on day-by-day retention even when the expected weekly return behavior is healthy.
Open Is Not Always A Return
If the retained action is a meaningful purchase, booking, use, or completion, a simple visit can inflate the retention story.
Common mistakes
Comparing retention before agreeing on the return event and expected return cadence.
Define the critical event and usage interval before reading the retention trend.
Q&A
Why does usage interval matter?
It tells you whether daily, weekly, monthly, or custom retention windows match the way customers should naturally return.
What should a retained user usually do?
They should return and perform the critical event that represents product value.
Read Cohorts Without Averages
Cohorts help show whether different customer groups retain, churn, or recover differently over time.
- Cohort analysis groups users by shared characteristics and tracks how those groups behave over time.
- Acquisition cohorts group users by when they first appeared, signed up, purchased, or installed.
- Behavioral cohorts group users by actions they performed or did not perform in a defined period.
In Practice
Cohorts Expose Hidden Differences
A flat average can hide that one signup month, channel, plan, or behavior group is retaining very differently.
Behavior Can Suggest A Hypothesis
If users who finish setup retain better, treat that as a hypothesis to investigate or test, not automatic proof of causation.
Common mistakes
Reading one blended retention number as if every customer segment behaved the same way.
Break retention into relevant cohorts before choosing a diagnosis or win-back action.
Q&A
When should you use an acquisition cohort?
Use it when the question is about groups that started in the same day, week, month, launch window, or signup period.
When should you use a behavioral cohort?
Use it when the question is whether a specific action, or not taking that action, is linked with later retention.
Separate Lifecycle Stages
Retention work changes when the customer is new, current, dormant, or returning after inactivity.
- New, current, and resurrected users make up active users in the lifecycle framework.
- A dormant user was active in the previous interval but did not use the product in the current interval.
- A resurrected user is active now after being inactive in the previous interval and active sometime before that.
In Practice
Growth Can Hide Churn
A rising active-user count can still be unhealthy if new users rise while current users fall and dormant users grow.
Win-Back Starts With Returned Users
When dormant users return, compare why they came back and whether they stay active after the return.
Common mistakes
Sending one retention campaign to every inactive or active customer stage.
Separate new, current, dormant, and resurrected users so the message and goal match the stage.
Q&A
Why break active users into lifecycle stages?
Different stages need different analysis and actions, so one active-user total is not enough.