Ever looked at a wall of user data and thought, “Now what?” You’re not alone. If you’re running a product, in growth, or just trying to figure out who’s actually using your app, you’ve probably heard about “cohorts.” It’s a fancy word for grouping users by real behavior, not just gut feelings or broad assumptions.
This guide is for anyone who wants to get past generic dashboards and actually see how different types of users act over time. We’ll walk through how to create and analyze user cohorts in PostHog, so you can stop guessing and start making decisions that matter. No fluff, no vague “insights” — just what works.
Why Cohorts? (And Why You Should Care)
Let’s get this out of the way: most analytics tools let you “segment” users. But if all you’re doing is filtering by “browser = Chrome,” you’re not really learning much. Cohorts let you group people by what they actually do — like users who finished onboarding, paid twice, or bailed after their first session.
Here’s what cohorts are actually good for:
- Finding patterns: See how real groups behave, like whether power users stick around or if new signups drop off.
- Testing changes: Did your new onboarding work? Compare cohorts before and after.
- Targeted actions: Send emails or trigger features only for users who meet certain criteria.
If you want to go beyond vanity metrics, cohorts are where you start.
Step 1: Set Up Your Events and Properties
Before you start slicing and dicing users, make sure your foundation isn’t made of sand. PostHog is only as good as the data you send it.
Checklist:
- Events: Are you tracking the actions that actually matter? (e.g. “Signed Up,” “Completed Purchase,” “Used Feature X”)
- Properties: Are they useful? (e.g. plan type, signup source, device)
- User identification: Are users consistently identified, even if they log in from multiple devices?
Pro tip: Don’t go overboard tracking every possible click. Focus on real milestones and behaviors that map to your goals.
If you’re not sure your tracking is solid, pause here and fix it. No amount of segmentation magic will help if your data is a mess.
Step 2: Create Your First Cohort in PostHog
Ready? Let’s make a cohort. Here’s how to do it without getting lost in the weeds.
-
Go to the “Cohorts” section
In PostHog, find “Cohorts” in the sidebar. If you don’t see it, make sure you’re on the latest version or have the right permissions. -
Click “New cohort”
You’ll see an option to create from scratch or based on an existing segment. Start from scratch for now. -
Define your criteria
This is where the magic happens. You can group users by: - Events (e.g., “users who completed checkout at least twice in the last 30 days”)
- Properties (e.g., “users on the Pro plan”)
- Behavioral combos (e.g., “users who signed up from a webinar and used Feature A”)
The builder lets you mix and match. Keep it simple at first: - Example: “Users who signed up in the last 7 days” - Example: “Users who triggered ‘Invite Friend’ at least once”
-
Name your cohort
Make it obvious — “Recent Signups,” not “Cohort 42.” -
Save and let it build
PostHog will crunch the numbers and show you who’s in the group. You can always edit later.
What to ignore: Don’t sweat the advanced filters or cohort math until you’ve nailed the basics. You can always get fancier later.
Step 3: Analyze Cohorts in Practice
Now you’ve got groups. Here’s how to actually use them.
3.1 Use Cohorts in Funnels
Funnels are where you see how users move (or drop off) between steps. Want to know if people who used your new feature finish onboarding faster? Add your cohort to the funnel as a filter.
- Compare different cohorts: Are your “Active Users” converting better than “New Signups”?
- Spot trouble early: If your “Activated Users” suddenly stop moving to the next step, something’s probably broken.
3.2 Retention Analysis with Cohorts
Retention charts answer, “Are people coming back?” Use cohorts to see if, say, users who came from a certain campaign stick around longer.
- Compare before/after: Did your product change help? Compare retention for users before and after launch.
- Look for surprises: Sometimes the “wrong” cohort (e.g., users from a niche source) ends up being your stickiest group.
3.3 Custom Dashboards
You can add cohort-based insights to dashboards — think “7-day retention by cohort” or “conversion rate for power users.” This is great for sharing with a team or execs who just want the highlights.
Step 4: Iterate and Refine (Don’t Set and Forget)
It’s tempting to make a bunch of cohorts and call it a day. Don’t. What matters is using what you learn to get better.
- Update cohorts: Your app changes, your users change, your cohorts should too.
- Kill useless ones: If a cohort never tells you anything new, delete it.
- Ask real questions: “Why are these users churning?” leads to better cohort definitions than “Let’s just group by geography.”
A/B testing tip: Cohorts are powerful for experiments. Test changes with a specific group, not your whole user base. You’ll get clearer results and fewer headaches.
Common Pitfalls (And How to Avoid Them)
Let’s be real — cohorts aren’t magic. Here’s what trips people up:
- Vague definitions: “Engaged Users” means nothing if you haven’t defined it. Be specific.
- Too many cohorts: It’s easy to drown in options. Start with 2-3 that matter, expand as needed.
- Stale data: If your cohort rules don’t update, you’re flying blind. Set them to auto-refresh if possible.
- Missing context: Cohorts are a tool, not the answer. Use them alongside user interviews, session replays, and gut checks.
A cohort is only as good as the question it answers. If you’re not clear on the question, you’re just making more work for yourself.
Real-World Examples (What Actually Works)
Here are cohorts that tend to be useful, no matter your product:
- New signups (last 7/30 days): How are fresh users behaving?
- Power users: People who hit your core feature X times.
- Churned users: Folks who haven’t come back in 30+ days.
- Upgraders: Users who moved from free to paid.
- Feature adopters: Users who tried the newest thing you shipped.
Skip cohorts like “users on Firefox in Canada.” Unless you have a burning reason, these tend to be trivia, not actionable.
Wrapping Up: Keep It Simple, Keep It Moving
Cohorts in PostHog can help you move from “I think…” to “I know…” about your users. But don’t let analysis paralyze you. Start with a question, define a group, look for patterns, and do something about it. Then adjust, repeat.
Segmentation isn’t about having the fanciest charts — it’s about making smarter decisions, a little faster, with a little more confidence. Keep it simple, keep iterating, and don’t be afraid to delete what’s not useful. That’s how you actually get value from your data — and your time.