You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@paolodamico and I spent an afternoon sifting through person data of various teams to understand how teams were using posthog.identify, specifically when email was provided, and if the problem with seemingly multiple copies of a person was a bug in implementation or an issue with how users were calling the function. We discovered several cases:
Flawless usage where persons are never duplicated against emails (example)
Duplicated persons where it seemed as though a company was calling posthog.identify on an incrementing ID which was causing multiples of a person. Here's an example
Duplicated persons across organization ids which is an expected side effect of signing up on different deployments explained here
Duplicated persons where the distinct_ids look like anonymous ids we generate. Here's an example of a person with multiple rows and the distinct_ids associated with them. Notice that none of the distinct_ids seem to be user passed in
*Duplicated means there are several person rows that all seem like they should be the same person based on their properties (email mainly)
Will continue adding findings/fixes to this thread
The text was updated successfully, but these errors were encountered:
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.
@paolodamico and I spent an afternoon sifting through person data of various teams to understand how teams were using
posthog.identify
, specifically when email was provided, and if the problem with seemingly multiple copies of a person was a bug in implementation or an issue with how users were calling the function. We discovered several cases:posthog.identify
on an incrementing ID which was causing multiples of a person. Here's an example*Duplicated means there are several person rows that all seem like they should be the same person based on their properties (email mainly)
Will continue adding findings/fixes to this thread
The text was updated successfully, but these errors were encountered: