There's currently no way to define a dynamic group of users based on event behavior with
time-window logic. For example, I want to find "users who did order_placed in the last 90 days
but NOT in the last 7 days" — a basic churn/lapsed-user cohort. This is a core feature in tools
like Mixpanel.
What I'd like
A Cohort Builder that lets you define user groups using conditions like:
- "Users who did [event] in the last [N] days" (with property filters)
- "Users who did not do [event] in the last [N] days"
- Combine multiple conditions with AND/OR logic
Once a cohort is defined, I should be able to:
1. View the list of matching profiles (with email, name, etc.)
2. Export as CSV for use in external tools (email campaigns, CRM, etc.)
3. Use the cohort as a filter/segment in other reports (e.g., "show retention for this cohort
only")
4. Cohorts should be dynamic — re-evaluated when viewed, not a static snapshot
Use case
I'm tracking order activity via order_placed events with a category property. I need to
identify churned traders (active in 90d, inactive in 7d), grouped by category, and export their
emails for re-engagement campaigns. Today, there's no
way to do it through the dashboard.
---
Sources:
- https://docs.mixpanel.com/guides/guides-by-topic/core-report…
- https://docs.mixpanel.com/docs/users/cohorts
- https://docs.mixpanel.com/docs/export-methods
- https://www.hotjar.com/mixpanel/glossary/cohorts/
- https://userpilot.com/blog/mixpanel-cohorts/