Cohort analysis is a way to group customers or users who share a meaningful characteristic and then compare how their behavior changes over time. In SaaS, teams often group customers by signup month, first purchase, plan, acquisition channel, or the date they completed activation. Following each group helps reveal patterns that an overall average can hide.
For example, total monthly active users may rise because the company is acquiring more users, even while each new group returns less often. Cohort analysis separates groups by when or how they started, making it easier to see whether retention, engagement, conversion, or revenue is improving for customers with similar starting points.
How Does Cohort Analysis Work?
A cohort is a group defined by a shared attribute or event. Teams choose the cohort definition based on the question they want to answer, then track the same group across consistent time intervals.
- Choose the question. Decide whether you want to understand retention, onboarding, conversion, revenue, or another behavior.
- Define the cohort. Group users by a meaningful event or attribute, such as the month they signed up or the channel that brought them in.
- Set the time intervals. Choose days, weeks, or months that match the product’s usage and purchase cycle.
- Choose the metric. Track a measure such as retained users, repeat usage, paid conversion, or revenue per cohort.
- Compare the groups. Look for changes across cohorts and over their lifecycle, taking account of incomplete or differently aged groups.
- Investigate and test. Use the pattern to form a hypothesis, validate it with qualitative feedback, and test a change where appropriate.
Types of Cohort Analysis
Acquisition cohorts
Acquisition cohorts group customers by when or how they entered the business. A SaaS company might compare users who signed up in January with those who signed up in February, or compare customers acquired through referrals with those acquired through paid campaigns. This can show whether retention or conversion differs by acquisition period or source.
Behavioral cohorts
Behavioral cohorts group users based on an action, such as completing onboarding, using a core feature, inviting a teammate, or reaching a first successful workflow. These groups can help teams investigate whether particular behaviors are associated with later retention or expansion. An association does not prove that the behavior alone caused the outcome.
Revenue cohorts
Revenue cohorts group customers by shared revenue characteristics, such as plan, contract type, or first paid month. Teams can compare recurring revenue, expansion, contraction, and churn over time. These views can help identify differences in customer economics and the long-term performance of acquisition channels.
Example of a Retention Cohort Table
Imagine a SaaS product groups new customers by the month they first subscribe and then measures the percentage who remain active in later months. A simplified illustrative table might look like this:
| Signup cohort | Month 0 | Month 1 | Month 2 | Month 3 |
|---|---|---|---|---|
| January | 100% | 82% | 74% | 68% |
| February | 100% | 86% | 78% | — |
| March | 100% | 88% | — | — |
These numbers are illustrative, not industry benchmarks or actual company results. The table suggests that later cohorts have higher retention in the observed months, but the February and March groups have not yet reached the same age as January. A fair comparison uses equivalent lifecycle periods and considers seasonality, product changes, customer mix, and data completeness.
Why Is Cohort Analysis Important for SaaS Businesses?
1. It makes retention trends easier to understand
Overall retention can change because the customer base is changing. Cohorts help show whether customers who joined more recently behave differently from older groups and whether the business is improving its ability to retain new customers.
2. It helps evaluate onboarding changes
If the team changes setup steps, education, or activation guidance, it can compare cohorts exposed to the new experience with earlier groups. The comparison can reveal useful patterns, although other changes during the same period may also affect the result.
3. It supports acquisition decisions
Different acquisition channels may attract customers with different needs and long-term behavior. Cohort analysis can help teams assess not only signup volume but also activation, retention, and revenue quality after acquisition.
4. It informs product prioritization
Behavioral cohorts can help teams explore whether users who adopt a core workflow retain better than those who do not. Product teams can use this as a hypothesis for research and experimentation rather than assuming the behavior automatically causes retention.
5. It improves revenue forecasting
Revenue cohorts can reveal patterns in expansion, contraction, and customer loss over time. These patterns can inform forecasts, but teams should account for differences in pricing, contract terms, customer segment, and the age of each cohort.
Cohort Analysis vs. Segmentation
| Approach | What it does | Example |
|---|---|---|
| Cohort analysis | Tracks groups over time based on a shared start point or behavior | Retention by signup month |
| Customer segmentation | Groups customers by shared attributes or needs | Retention differences between small businesses and enterprise accounts |
| Funnel analysis | Measures progression through a sequence of steps | Signup to activation to paid conversion |
These methods can work together. A team might segment customers by plan, then compare retention cohorts within each segment. It can also use funnel analysis to understand where users drop off and cohort analysis to see whether the people who pass a step remain active over time.
Best Practices for Reliable Cohort Analysis
Define the cohort event and metric before looking at the results. For retention, specify what counts as active, whether the measure is based on login or meaningful product use, and how cancellations or paused accounts are handled. Keep definitions stable when comparing periods.
Use time windows that match the product’s normal usage cycle. Daily cohorts may be useful for a frequently used tool, while monthly intervals may be more appropriate for products used around billing or reporting cycles. Avoid comparing groups with different observation periods without making that limitation clear.
Check sample size and customer mix. Small cohorts can fluctuate because a few customers change behavior, while changes in pricing or target audience can make a newer cohort different from an older one. Pair quantitative analysis with interviews, support themes, and product feedback to understand why a pattern may exist.
Finally, turn findings into a testable question. If customers who complete a setup step retain better, investigate whether the step helps them achieve value or simply attracts more motivated users. Test a focused onboarding change and measure its effect instead of assuming correlation proves causation.
Common Cohort Analysis Mistakes
- Comparing cohorts at different ages: Newer groups have had less time to churn, renew, or expand.
- Changing metric definitions: A change in what counts as active can create an apparent improvement that is not real.
- Using very small groups: A few accounts can dominate the result and make it unstable.
- Assuming causation: A cohort difference can reflect customer mix or other changes, not only the event being studied.
- Ignoring incomplete data: Tracking gaps, delayed events, and product instrumentation changes can distort comparisons.
- Reporting results without action: Analysis is most useful when it informs a decision, experiment, or further investigation.
Frequently Asked Questions
What is the difference between a cohort and a segment?
A cohort is commonly followed over time and often shares a starting event or period. A segment groups people by shared attributes or characteristics. The same group can be both a cohort and a segment, depending on how it is defined and used.
Which metrics can be analyzed by cohort?
Teams can analyze retention, churn, activation, conversion, feature adoption, revenue per customer, expansion, support usage, and other measures, provided the metric is defined consistently and the data is available.
How often should SaaS teams run cohort analysis?
The cadence depends on the product’s usage cycle and the decision being made. Teams may review cohorts monthly or quarterly, while fast-moving product experiments may require more frequent analysis. Ensure each cohort has enough time and observations for a useful comparison.
Can cohort analysis prove that a product change worked?
It can show whether outcomes changed across groups, but a simple before-and-after cohort comparison may be affected by other factors. Controlled experiments or additional analysis can provide stronger evidence about causation.