What Is a Cohort and How to Run a Cohort Analysis

A headline metric like "we have 10,000 customers" says little about the real health of your business. Are the ones who arrived in January still with you? Did March's campaign customers stick around more or less than June's? Answering that is exactly what cohort analysis is for. This guide covers what a cohort is, how the table is built, and what decisions it lets you make.

What a cohort is

A cohort is a group of users or customers who share a common characteristic within the same time period. The most common one is the acquisition cohort: everyone who signed up or made their first purchase in the same month.

For example, "the January 2026 cohort" is every customer who made their first purchase in January. From there, you follow that group month by month to see how many stay active.

What cohort analysis is

Cohort analysis means grouping your customers into cohorts and measuring their behavior over time, instead of looking at a single overall average. Its big advantage: it strips out the effect of time and shows whether your product retains better or worse with each new generation of customers.

Types of cohorts

• Acquisition cohorts: grouped by when the customer started (the most common).
• Behavioral cohorts: grouped by an action they took (used a feature, bought a certain product).

How to read a cohort table

The cohort retention table is a matrix. Picture this:

• Each row is a cohort (the month the customers arrived).
• Each column is the time elapsed since then (month 0, month 1, etc.).
• The values are the percentage still active at each point.

Reading the example: the March cohort retains better (70% in month 1) than January's (62%). That suggests something improved, perhaps better onboarding or a more polished product.

What cohort analysis reveals

1. Whether retention is improving: compare newer cohorts against older ones.
2. Where you lose customers: if everyone drops sharply in month 1, your onboarding is failing.
3. The impact of a change: you shipped an improvement in March; the March cohort reflects it.
4. Real LTV: tracking spend by cohort estimates how much each customer is worth over time.

An applied example

A subscription company notices its total customer count is growing but revenue is flat. A cohort analysis reveals that recent cohorts drop off before month 3. The apparent growth came from acquiring a lot, not from retaining. The company refocuses effort on the first-30-day onboarding, and two quarters later new cohorts retain 15 points higher at month 3.

The role of conversations in retention

Much of what decides whether a cohort stays or leaves happens in the first few interactions: the welcome, resolving the first question, timely follow-up. When those conversations are scattered across WhatsApp, email, and Instagram, a new customer can easily fall through at the critical moment. With an omnichannel platform like Omnifox you can automate onboarding messages, follow-ups, and re-activations from a single inbox, so you can act on cohorts that start to drop before you lose them.

How to start your own analysis

1. Define the event that marks the cohort (first purchase, sign-up).
2. Group customers by the month of that event.
3. Measure a key metric (active customers, revenue) in each later period.
4. Build the table and look for patterns across rows.
5. Act: if month 1 is weak, work on onboarding; if the drop is late, review ongoing value.

Retention cohorts vs revenue cohorts

So far we've talked about retention (how many customers stay active), but you can build cohorts on almost any metric. A very revealing variant is the revenue cohort, where instead of the percentage of active customers you measure how much money each cohort generates over time. Sometimes you lose customers but the ones who remain spend more, and the revenue cohort grows even as the retention cohort declines. That phenomenon, known as net revenue retention, is one of the strongest signals of a healthy business.

Combining both views gives you a complete diagnosis: the retention cohort tells you how many stay, and the revenue cohort tells you how much the ones who stay are worth. Together they reveal whether your growth is solid or whether you depend on constant acquisition to paper over the leak.

Conclusion

A cohort groups customers by when they started, and cohort analysis lets you see real retention separated from the effect of time. It's one of the most honest tools for knowing whether your business is genuinely improving or just growing by acquiring more. Read it across rows, spot where customers drop, and act on those moments, especially onboarding.

Want to take better care of every cohort from the very first conversation? Try Omnifox and automate your customers' onboarding and follow-up.