Allow event property filters when creating a cohort

[joelmatch]

Is your feature request related to a problem?

*No. This is an enhancement to make it less clunky to achieve a simply cohort comparison. I can get by if you ignore this, but the pattern exists elsewhere in the product and it annoys me that it's not followed in cohort building.

• I want to breakdown a funnel by two cohorts
• Cohorts should be based on performing an event with a property
• You can't use event properties to build cohorts currently
• This is a feature I am used to in Amplitude

My specific example:

• I'm trying to determine if we should be investing eng resources into a better login auth process than our current homegrown passwordless login.
• We understand ~24% of users experience a login failure but 93% eventually login.
• I want to understand if there are conversion differences AFTER the login expiry between the two groups to help make a case to invest in reducing the error rate

Describe the solution you'd like

• When creating a cohort
• Matching users who performed an event
• Will see an option to add a filter based on properties
• For example: performed event login with property result = failure

Describe alternatives you've considered

• Creating an "action" that uses the custom event + event property filter, then using these actions to create the cohort. It works but adds an extra step and new "action" that permanently shows on the event selection dropdown everywhere
• Currently using Amplitude to achieve this easily but we are evaluating moving to PostHog
• Started with 2 funnel charts side by side in a dashboard, but not as easy for direct comparison

Additional context

Thank you for your feature request – we love each and every one!

THANK YOU POSTHOG! I am so onboard with your philosophy and want to see you succeed.

[macobo]

Thanks for the suggestion @joelmatch - cleaned up the ticket a tiny bit. :)

[posthog-bot]

This issue hasn't seen activity in two years! If you want to keep it open, post a comment or remove the stale label – otherwise this will be closed in two weeks.

[neilkakkar]

still valid