N-Day Retention
The percentage of a starting cohort that returns on a specified day or period number after its initial event.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/n-day-retention/
01What it means
N-Day Retention measures whether cohort members perform a defined return event on Day N, Week N, or another fixed interval after entering the cohort. It is distinct from unbounded or return-on-or-after retention, which counts later returns too.[1][2]
- Cohort entry
- Defined event and entry date or period
- Day N
- One fixed interval after entry
- Return event
- Explicit qualifying behavior
- Eligible denominator
- Cohort members observable through Day N
State cohort entry, return event, time-zone and interval convention; unbounded retention answers a different question.
02Why it matters
A fixed interval creates a clear cohort comparison and can expose where recurring use decays after activation.
03When to use it
Use it when the product has a meaningful expected cadence and define calendar versus rolling windows explicitly.
04How it works
- Define cohort-entry and return events.
- Choose the interval and time-window convention.
- Group users by entry date or period.
- Divide returning cohort members by eligible cohort members.
- Compare curves and segments with confidence and context.
05A practical example
A weekly planning tool measures Week-4 retention as the share of newly activated teams completing a planning session in their fourth week.
06Where it breaks
Fixed windows can miss users who return just outside the boundary, and immature cohorts create incomplete observations.
Common mistakes
- Mixing return-on and return-on-or-after methods.
- Comparing cohorts before all reach Day N.
- Choosing N without regard to natural usage frequency.
07Sources
- Build a retention analysisAmplitude Docs · Official documentationBuild a retention analysis
- Interpret your retention analysisAmplitude Docs · Official documentationInterpret your retention analysis