You’re staring at PostHog, trying to answer: “How many of our enterprise customers are actually using the product each week?”

You could build a cohort. You could set up groups. You could filter by properties. But which layer of PostHog data actually answers the question?

Most teams guess wrong—then wonder why their dashboards show inflated numbers or miss key segments entirely.

I’m JJ from Vision Labs. We wire PostHog for software teams who need accurate utilization, conversion, and retention metrics. Here’s how to pick the right tool (cohorts, groups, or plain event filters) for each business question.

The PostHog data hierarchy: properties → events → people → cohorts + groups

Think of PostHog data in layers. Each layer rolls up into the next.

Properties sit inside events:

• Which page someone viewed
• Browser type (Safari, Chrome)
• Device (mobile, desktop)
• URL path or slug

Events belong to people:

• Page view + added to cart + newsletter signup + account creation + trial start
• Each person generates multiple events

People become cohorts:

• Cohorts group people by shared characteristics or behaviors
• Anyone who completed a quiz in the last 30 days
• Users who viewed pricing but never purchased
• Pro members who’ve watched 5+ videos

People also roll into groups:

• Groups organize people by billing structure or organization
• 10-seat enterprise plan with multiple users
• Company-level usage across all accounts

The rule: properties feed events, events belong to people, people populate cohorts and groups.

Cohorts vs groups: when to use each

Cohorts = flexible segmentation you define with filters.

Use cohorts when you want to:

• Identify users who performed (or didn’t perform) specific actions
• Build retargeting lists based on behavior
• Track conversion through funnel stages
• Segment by engagement level

Example: “Anyone who viewed cart and then purchased within 15 days.”

You define the criteria. PostHog updates the membership automatically as users match or stop matching your filters.

Groups = fixed organizational structures for multi-user accounts.

Use groups when you:

• Offer team or enterprise plans with multiple seats
• Need to measure organization-level utilization
• Want to see which companies are active (not just which individuals)
• Track usage across accounts that share billing

Example: 100 organizations with group plans. How many have at least one active user each week? How much total volume are they using?

Groups require implementation. Every event must carry the group identifier—or the rollup breaks. It’s an add-on feature (first million group events free, then $0.000007 per event after 2M).

The implementation rule for groups: every event needs the group ID. Unlike people (where you identify once), groups only work if you send the group ID with each event. Miss it, and you lose the organizational rollup.

How to create a cohort in PostHog

Hit Cohorts in the sidebar. Click New cohort.

Example 1: Anyone who completed a quiz in the last 30 days

• Name: “Quiz completions”
• Criteria: Completed event quiz after 30 days
• Save

PostHog loads a list of ~1,600 people who match. Each row shows the person ID. Click any row to see their full profile and event stream.

Example 2: Sequence-based cohort (cart view → purchase)

• Criteria: Completed sequence of events
• First event: cart viewed
• Then: purchased
• Timeframe: within 15 days
• Save

Now you have a cohort of people who viewed cart and bought within 15 days. Use this to:

• Retarget similar users who viewed cart but haven’t purchased yet
• Measure conversion rate from cart to purchase
• Export the list (CSV) for CRM upload

Cohort trigger example: video engagement → CRM sync

One customer tracks video watch time. Anyone who watches more than 5 minutes of any video gets added to a cohort. That cohort triggers a webhook to their CRM. Result: high-intent leads auto-populate their sales funnel.

Where cohorts show up in your analytics

Cohorts aren’t just lists—they filter every PostHog dashboard, insight, and funnel.

Example: Conversion rate filtered by quiz completers

• Open any dashboard (e.g., “Pricing to Purchase”)
• Add filter: Cohorts → select Quiz completions
• Apply

You’ll see conversion rate for people who completed a quiz vs the full user base. In one customer’s data, users who completed a quiz had remarkably high purchase rates—so they started pushing more users into quizzes.

The timing trap: Cohorts are user-based, not event-sequenced. If someone purchased and then completed a quiz, they still appear in the “quiz completions” cohort. The cohort includes anyone who ever matched the criteria—regardless of event order.

To enforce sequence, use funnel insights with time constraints instead of cohort filters.

Export cohorts for retargeting or CRM upload

Click the three-dot menu on any cohort. Select Export all columns.

You’ll get a CSV with:

• Person ID
• Email (if identified)
• Custom properties

Upload to your CRM, ad platform, or email tool.

When to use group analytics (and when to skip it)

Groups cost extra. First million group event calls are free; after that, it’s $0.000007 per event (on top of your standard event volume).

Use groups if:

• You have multi-seat plans (teams, enterprise)
• You need company-level utilization metrics
• Different people log in under the same organization
• You want to measure “active organizations” not just “active users”

Skip groups if:

• You only have individual accounts
• You can answer your questions with cohorts
• You don’t have the dev resources to send group IDs with every event

We’ve worked with teams who tried shortcuts—like identifying a group once, then assuming PostHog would associate all future events with that group. It doesn’t work that way. Every event needs the group ID in the payload, or the rollup fails.

Questions we answer with cohorts and groups

Cohorts:

• Which users completed onboarding but never activated a key feature?
• Who viewed pricing in the last 7 days but didn’t upgrade?
• How many people watched more than 10 minutes of content this month?
• Which trial users logged in 3+ times in their first week?

Groups:

• How many organizations had at least one active user this week?
• Which companies are using less than 20% of their seat capacity?
• What’s the average event volume per organization?
• Which enterprise accounts haven’t logged in for 30 days?

Practical implementation checklist

Before you build cohorts or groups:

• Confirm your events fire consistently (check via Live events in PostHog)
• Verify properties populate correctly (browser, page path, custom fields)
• Test one simple cohort (e.g., “anyone who did X in last 7 days”)
• Check cohort membership updates as users perform actions
• If using groups: ensure every event payload includes group ID
• Test group rollup by filtering a dashboard by a known organization

Timeline: Cohorts work immediately once events are tracked. Groups require backend changes to attach group IDs to every event budget 1–2 sprint cycles for implementation and QA.

Next steps

If you’re using PostHog but your data still feels scattered:

1. Map your key user journeys to events and properties
2. Build 3–5 cohorts around funnel stages (viewed pricing, started trial, activated feature, upgraded)
3. Filter one dashboard by each cohort to verify conversion rates
4. Decide if you need group analytics—or if cohorts solve your segmentation needs

Need a PostHog audit? We help teams implement event tracking, cohort strategies, and group analytics correctly. Go to visionlabs.com/contact and mention PostHog for a free audit we’ll review your setup and show you where data is breaking or where you can get more value from features you’re already paying for.