This message was deleted.

In our application the flow goes: • User lands on page, gets anonymous_id • User gets assigned to A/B based on anon_id • User creates an account and gets a proper user_id • User completes an action several days later - at this point their anonymous_id has likely changed Is there a way I can coach growthbook to look for the user_id that gets seen with an anonymous_id at the time of signup? And from then onwards, our experiment metrics should look for that user_id and NOT the anonymous_id? Is this making sense?

When you setup your datasource i.e. BigQuery you can select multiple identifier types i.e. anon id and loggein visitor. And there you can also select to join these two so they are considered as one group during analysis. Hope this helps you.

I'll check it out. Thank you! My concern with the join tables functionality is we've got a many to many relationship between anonymous_ids and user_ids. Multiple users will go through the account creation process on the same shared device/browser and thus have the same anonymous_id. And the same user may use multiple devices and thus have multiple anonymous_ids. This may be the best we can do though. Thanks again