N-day Retention
AdvancedN-day retention is the share of a cohort that performed the return action exactly on day N after their start day, not earlier and not later.
This is the strictest way to measure retention, so it produces the lowest numbers and the most jagged curve; a product with a weekly rhythm peaks on days 7 and 14 and dips in between. It differs from rolling or unbounded retention, where a return on day N or on any later day also counts, which yields a higher figure and a smoother line. The most common mistake is comparing an N-day figure from one tool with a rolling figure from another; the same product can report 12 percent under one definition and 30 percent under the other. A second detail is the day boundary: some tools use the calendar day in the project timezone and others a 24-hour window from the start moment.
How it is calculated
N-day retention = cohort users who performed the return action exactly on day N ÷ cohort size on day zero × 100
The denominator is the cohort size on day zero and is never re-based; a user who joins mid-range belongs to a different cohort. Day zero is the day of the qualifying start action, so day 1 means the day after it. The unit is a percentage, and only cohorts whose day N has fully elapsed should enter the calculation, otherwise immature cohorts drag the figure down artificially.