Is there a ‘best practice’ way of creating a metric like subscription retention for a test, given that users can have different renewal dates (i.e. the calculation should only include users for which their renewal date has passed, and they either renewed or they didn’t)? The same logic applies to different metrics where the users to include in a denominator depend on some externally defined date, specific to the user.

Hi Callum you could calculate MRR or some kind of LTV equation