What is cohort analysis?
Cohort analysis is the practice of grouping members by when they joined and following each group separately over time, rather than reporting one blended average across everybody currently in the business.
Averages hide the thing you need
A blended retention or engagement figure mixes members who joined last month with members who joined four years ago. The two behave nothing alike, so the average moves for reasons nobody can explain and nobody can act on.
Split by join month and the picture changes completely. You can see that members who joined after a particular change stayed longer, or that every cohort drops sharply in the same week, which is a product problem rather than a member problem.
It is how you prove a change worked
Without cohorts there is no honest way to attribute an improvement. Overall numbers drift with seasonality, marketing and mix. With cohorts, you compare the people who joined before a change against the people who joined after, and the comparison means something.
Cohort analysis matters most for exactly the decisions operators find hardest to justify, such as whether a platform investment actually improved retention.
You do not need a data team
A spreadsheet works. Join month down the side, months since joining across the top, percentage still active in each cell. Reading down a column shows whether newer cohorts behave better than older ones. Reading across a row shows where in the lifecycle people leave.
In most fitness businesses the answer to the second question is the first eight weeks, which immediately tells you where to spend.
Questions operators ask
Four that come up on almost every call about cohort analysis.
Usually everybody who joined in the same month, though it can be anybody who shares a starting condition such as a campaign or a club. The defining feature is that you follow the group over time instead of blending it into the base.
Because the overall figure changes with mix, seasonality and marketing, which makes it impossible to attribute anything. Cohorts isolate the effect of what you changed by comparing people who joined before it with people who joined after.
At least twelve months for retention, since annual contracts and seasonal patterns distort anything shorter. For activation and early engagement, the first eight weeks carry most of the information you need.
No. A spreadsheet with join month down one axis and months since joining across the other is enough to answer the important questions. Dedicated tooling helps at scale but is not what stops most operators from doing this.
Building a fitness platform?
Tell us where you are and what your members are doing. We will tell you what we would do about it.
Start a conversationRelated terms
The terms that come up in the same conversations as cohort analysis.
Member Churn
Member churn is the rate at which members leave over a period, the inverse of retention, and the number that quietly sets the ceiling on how large a...
Running clubsMember Retention
Member retention is the share of members who remain with a business over a given period, usually measured monthly or annually, and in fitness it is...
Knowing if it worksActivation Rate
Activation rate is the share of new members who reach a defined first useful moment in the app, such as completing a first logged session or...