When you evaluate business metrics, you need to understand how they evolve over time for a specific group of users. This technique is called cohort analysis. It helps you track the user lifecycle in your product and spot trends that averages would hide.
Why You Need Cohort Analysis
Let me start with an example. Someone might ask you about the ARPU (Average Revenue Per User) of your product. You probably know the answer right away. However, that single number can be misleading.
The ARPU of a user in their first month is not the same as in their tenth month. As a result, a simple average blends new and old users together. This hides important patterns in your data.
A cohort tracks how a metric changes for a specific group of users over their lifetime. You can measure cohorts in any time unit: days, weeks, months, or years. Similarly, you can apply cohort analysis to any metric: retention, ARPU, conversion, churn, and more.
A Practical Example: ARPU Cohorts
Consider the following table. It shows ARPU by monthly cohorts from January to June.
| ARPU | Month 0 | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 |
|---|---|---|---|---|---|---|
| January | $0.07 | $0.20 | $0.30 | $0.10 | $0.05 | $0.02 |
| February | $0.08 | $0.20 | $0.30 | $0.10 | $0.05 | |
| March | $0.18 | $0.28 | $0.40 | $0.20 | ||
| April | $0.10 | $0.20 | $0.30 | |||
| May | $0.11 | $0.20 | ||||
| June | $0.12 | |||||
| Average | $0.11 | $0.22 | $0.33 | $0.13 | $0.05 | $0.02 |
This table reveals a lot of information. For instance, users who signed up in January had an ARPU of $0.07 in Month 0. By Month 2, their ARPU grew to $0.30. In this way, you get the average ARPU from the registration date through each subsequent month.
Reading Columns: Marketing and Product Effectiveness
When you analyze the columns, you measure how effective your marketing and product have been over time. Take Month 0 as an example. The ARPU improves steadily from January to June, with a notable spike in March.
This could mean two things. Either you improved the first-month user experience and boosted conversion, or you refined your marketing strategy to acquire higher-quality users. Both are valuable insights that a simple average would miss.
Reading Rows: The User Lifecycle
On the other hand, when you analyze the rows, you measure the user lifecycle. Here you can see that users tend to spend more in Months 2 and 3. After that, revenue drops sharply.
This drop could signal that users lose interest after their third month. Alternatively, it might point to a product design issue that causes users to churn. Either way, cohort analysis makes the problem visible.
Spotting Outliers: The March Cohort
To understand the full power of cohort analysis, let us compare the March cohort against the overall average.
| ARPU | Month 0 | Month 1 | Month 2 | Month 3 |
|---|---|---|---|---|
| March | $0.18 | $0.28 | $0.40 | $0.20 |
| Average | $0.11 | $0.22 | $0.33 | $0.13 |
Clearly, March outperforms every other month. Your team should investigate what happened. Most likely, a higher-quality acquisition campaign ran that month. Without cohort analysis, this insight would stay hidden in the overall average.
Important note: Always check the sample size. If March had 100 active users while other months had 1,000, the higher ARPU might simply reflect a smaller denominator. For this reason, include a column with absolute values alongside your cohort percentages.
Cohort Analysis for Retention
Now let us apply cohort analysis to monthly retention. The table below shows retention cohorts from January to June.
| Retention | Month 0 | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 |
|---|---|---|---|---|---|---|
| January | 50% | 30% | 20% | 15% | 20% | 15% |
| February | 55% | 35% | 23% | 18% | 15% | |
| March | 50% | 28% | 17% | 14% | ||
| April | 45% | 30% | 19% | |||
| May | 60% | 50% | ||||
| June | 50% |
The chart above helps you draw three important conclusions quickly:
- February stands out. Its retention curve sits above all others. What made those users more engaged? Were they higher-quality acquisitions?
- May shows great early promise. Retention in the first months is significantly higher. Did a product change improve the experience, or did a new marketing strategy attract better users?
- January spikes in Month 4. Users came back after months of decline. Did a remarketing campaign re-engage them?
Key Takeaways
In summary, cohort analysis gives you a much clearer picture of your business metrics. It reveals patterns in the user lifecycle that simple averages cannot show.
You can run cohort analysis on both absolute metrics (revenue, active users) and relative metrics (retention, churn, conversion). Furthermore, you can segment your cohorts by market, product, or acquisition channel for deeper insights.
The most practical business application is measuring marketing ROI through cohorts. This lets you understand exactly how much value each acquisition campaign generates over time. We will explore that topic in a future article.
s steadily from January to June, except for a spike in March.This could mean two things. Either you improved the first-month user experience and boosted conversion, or you refined your marketing strategy to acquire higher-quality users. Both