Your dashboard says you have 12,000 active users this month, up from 11,000 last month. Good news? You can't tell. That one number hides whether you're keeping the users you already had or just replacing the ones who left.
Cohort analysis is how you find out. It groups users by something they share, such as the week they signed up or an action they took. Then it tracks each group over time. Instead of one blurry total, you see who sticks around, who leaves, and when.
This guide explains what cohort analysis is, how to read a retention curve, how to run one step by step, and when it can fool you.
What is cohort analysis?
A cohort is a group of users who share a starting point. Everyone who signed up in the same week is one cohort. Everyone who placed their first order in March is another. Cohort analysis tracks each group on its own as time passes, so you can compare groups instead of mashing them into one average.
Picture a spreadsheet grid. Each row is a cohort. Each column is a period after the start: week 1, week 2, week 3 and so on. Each cell shows how many people in that cohort were still active in that period. Older cohorts have had more time, so the filled-in cells form a triangle.
Here's why this matters. Say overall retention looks flat at 30%. That could mean every cohort keeps a steady 30%. Or it could mean your old cohorts are collapsing while new ones keep 45%. One is healthy and one is on fire, and a single number can't tell them apart. A cohort grid can.
Acquisition vs behavioral cohorts
There are two ways to group users. They answer different questions.
Acquisition cohorts
Acquisition cohorts group users by when they started: signup week, first-purchase month or install date. This is the most common kind. It answers one question well: are newer users sticking around better or worse than older ones?
Say you shipped a new onboarding flow in April. Compare the March cohorts with the April cohorts. If April keeps more people four weeks in, the new flow worked. Acquisition cohorts turn "we changed something" into "we changed something and retention went up 9 points."
Behavioral cohorts
Behavioral cohorts group users by what they did: created a dashboard, invited a teammate, or hit some key step in their first week. These answer a more useful question: which actions predict that people stay?
Say users who connect an integration in week one stay at 60%, and everyone else stays at 25%. You've found your activation lever. Now you have a reason to guide every new user to that action.
| Acquisition cohort | Behavioral cohort | |
|---|---|---|
| Grouped by | When users started (signup week or month) | An action they took (used a feature, hit a key step) |
| Answers | Are newer users staying better than older ones? | Do users who do X stay better than those who don't? |
| Example | Everyone who signed up in March | Everyone who created a dashboard in week 1 |
| Best for | Checking whether changes improved retention over time | Finding the actions that keep people around |
Most teams start with acquisition cohorts to track the trend. Then they add behavioral cohorts to find the levers behind it.
How to read a cohort table
This is where people get stuck, so let's read one. Below is weekly retention for four signup cohorts (example data). Week 0 is the signup week, so it is always 100%. Each later column is the share of that cohort who came back and did something meaningful that week.
| Signup week | Users | Week 0 | Week 1 | Week 2 | Week 3 | Week 4 |
|---|---|---|---|---|---|---|
| Mar 3 | 480 | 100% | 41% | 33% | 29% | 27% |
| Mar 10 | 510 | 100% | 39% | 31% | 28% | 26% |
| Mar 17 | 465 | 100% | 44% | 36% | 33% | 31% |
| Mar 24 | 530 | 100% | 47% | 40% | 36% | 34% |
Read it two ways.
Across a row you see the retention curve for one cohort. The Mar 3 group drops from 100% to 41% in one week. Then the fall slows down: 33%, 29%, 27%. That early drop is normal. Most products lose a big share of signups right away. What matters is whether the curve flattens. A flat curve means you have a core of users who keep coming back. A curve that keeps sliding toward zero means a leaky bucket.
Down a column you compare cohorts at the same age. Look at Week 4: 27%, 26%, 31%, 34%. Later cohorts stay better. Something you did in mid-to-late March raised four-week retention by about 8 points. Reading down the columns is the most useful thing a cohort grid gives you, and you can't see it in a total-active-users chart.
How to run a cohort analysis, step by step
You don't need SQL or a spreadsheet full of VLOOKUPs. Any product analytics tool with a retention report builds the grid for you. Here's the process.
- Identify your users. Cohort analysis needs to know the same person across visits. So you call an identify method when someone logs in, and events tie to a stable user ID instead of an anonymous visit. See identifying users for how this works.
- Pick the cohort definition. Start with an acquisition cohort (signup date) to get your baseline trend. Once that works, add a behavioral cohort around an action you think matters.
- Pick the return event. Decide what "retained" means. A login doesn't say much. Pick an event that shows real value: viewed a report, sent a message, placed an order. Retention is only as honest as the event you choose.
- Choose the interval. Daily for apps people use many times a day. Weekly for most B2B and SaaS products. Monthly for tools people use now and then, like tax software. Match the interval to how often a happy user should come back.
- Read the grid, then the curve. Scan down the columns for the cohort-over-cohort trend. Then read across the rows to see where the curve flattens. Build the report once in Pieowl's retention report and it keeps updating as new cohorts age.
A worked example
Here is an example. A small SaaS team is worried about churn. Their total active users are flat, and they read that as "stable." The cohort grid tells a different story.
Their March signup cohorts keep about 22% at week four. In early April they ship an onboarding checklist that nudges new users to create their first dashboard. The April cohorts come in at 31% at week four. That is a 9-point lift, and the flat top-line number hid it.
Then they build a behavioral cohort: users who created a dashboard in week one versus those who didn't. The dashboard creators stay at 48%; everyone else at 19%. Now they have proof, not a hunch. Creating a dashboard is the activation moment, so next quarter's work is clear: get more new users there, faster.
This is the difference between cohorts and a conversion funnel. A funnel shows where users drop off on the way to a goal, in one short window. A cohort shows whether they keep coming back weeks later. You need both. The funnel finds the leak in signup. The cohort tells you whether the people who got through actually stay.
When cohort analysis misleads
Cohort grids look official, which makes their weak spots risky. Watch for these.
- Small cohorts are noise. A cohort of 30 users swinging from 20% to 40% is probably chance, not a trend. Don't act on small cohorts unless the effect is huge.
- Seasons fake trends. A cohort that signed up the week before a holiday may look worse than a normal week, for reasons that have nothing to do with your product. Compare like with like.
- Vanity return events. If "retained" means "opened the app," retention will look great and mean nothing. Tie retention to an event that shows value.
- Changing the definition midway. If you change the return event or the key action halfway through, old and new cohorts stop being comparable. Lock your definitions before you read trends.
- Only looking at who stayed. The grid shows who stayed. To understand churn, also look at the users who left: what they did, or didn't do, before they went quiet.
Cohorts, segments, and funnels
These three get mixed up all the time. The short version:
- A funnel measures steps toward a goal over a short window (view → add to cart → checkout). It answers where do people drop off?
- A cohort follows one group over a long window. It answers do people keep coming back?
- A segment is a filter you can apply to either one, like "users in Germany" or "mobile signups." It answers which slice are we looking at?
They work together. Split a cohort by where users came from and you may find that paid signups stay far less than organic ones. That kind of insight changes where you spend money.
Track retention without cookies
Pieowl has retention and cohorts built in, next to funnels and event tracking. You can define a cohort by what people did (the events they triggered) or by their profile properties. Then filter your reports to people "in cohort" and compare. Read the retention curve and find your activation lever.
You can also ask your own AI assistant, like Claude or Cursor, through Pieowl's MCP server, for example "In which week do most new users stop coming back?"
Pieowl sets no cookies, so in many cases you don't need a consent banner for analytics. See when you still need one. Your data is stored in the EU. Start with a 14-day free trial, no card needed. Plans start at $9 a month. See pricing.
What is cohort analysis?
Cohort analysis groups users by a shared starting point, like the week they signed up or an action they took, and tracks each group over time. Instead of one average that hides what's happening, you see whether each group keeps coming back. It's the clearest way to measure retention and spot churn early.
What's the difference between acquisition and behavioral cohorts?
Acquisition cohorts group users by when they started (signup week or month). They answer "are newer users staying better than older ones?" Behavioral cohorts group users by what they did (used a feature, hit a key step). They answer "which actions predict that people stay?" Acquisition cohorts track the trend; behavioral cohorts find the lever behind it. Most teams use both.
How do you do a cohort analysis?
Identify your users so events tie to a stable ID. Pick a cohort definition (start with signup date). Choose a return event that shows real value, not just a login. Pick an interval: daily, weekly or monthly. Then read the grid: down the columns for the cohort-over-cohort trend, and across the rows to see where the curve flattens. A product analytics tool with a retention report builds the grid for you, with no SQL or spreadsheets.
What's the difference between cohort analysis and segmentation?
A cohort is a group with a shared starting point, tracked over time. It answers "do these users keep coming back?" A segment is a filter on your data, like "users in Germany" or "mobile signups." It answers "which slice am I looking at?" They work well together: split a cohort by where users came from and you'll often find paid and organic signups stay very differently.
What is the best tool for cohort analysis?
Mixpanel and Amplitude are well-known product analytics tools with strong cohort and retention reports, and both have free plans for smaller volumes. Pieowl lets you build cohorts from events or profile properties, filter reports by cohort, and ask your AI assistant about them over MCP. It sets no cookies and has a 14-day free trial. Whatever you pick, choose a tool that ties retention to a meaningful return event and lets you build behavioral cohorts, not just signup-date ones.
Can Google Analytics do cohort analysis?
Yes. GA4 has a cohort exploration in its Explorations section. You can group users by their first visit, any event, a transaction, a conversion, or a specific event you choose, and view retention daily, weekly or monthly. It works well for a quick look. In Pieowl, you save a cohort once, reuse it as a filter across reports, and open the profile of any user in it.