I ran an event export and got way more events than what Mixpanel reports show... looking at the export a bit more carefully, I realized that a lot of the events were duplicated, some of them up to 5 times... I checked all the fields too, they're all identical... does anyone know what is happening?
Hey Vlad! If Mixpanel received an event two times due to network issues/retries, the UI will deduplicate the event at query time, known as Query Time Deduplication. However, the Raw Export API exports the raw data as it is currently stored in Mixpanel. The behavior seen indicates that there are duplicates of events being sent to Mixpanel. You should be able to deduplicate the data based on the $insert_id of the events.
wouldn't an event that was received twice have slightly different timestamps, at least in one of the many we keep track of?
like they'll have the same time field but maybe a different processed_at or received_at field, right?
I'm seeing a lot of perfectly duplicated events
If the duplicate events are getting triggered at the same time, it is possible for those values to be the same as well. Is the behavior seen for specific events? If you go through the same user flow, are you able to replicate the behavior to confirm if there are multiple payloads being sent to Mixpanel?
I had di export of about 100 events, most of them were duplicated exactly, some up to six times
Are you able to replicate the behavior? The duplicates do indicate that multiple payloads that are the same are being sent.
you mean can I export them again and see if they're duplicated? I haven't tried, but I'll give it a go with a different event where I have a handful of them I need to fix
The behavior seen indicates that the same event payload is being sent to Mixpanel multiple times when an event is fired. Are you able to replicate the behavior as a user where you see multiple payloads being sent to Mixpanel? This would be on the implementation side, not just on the export side.
oh, I see what you mean... we route our events through Segment, and as far as I can tell those events don't show up multiple times in Segment... if there's duplication it's somewhere between Segment and Mixpanel most likely
Interesting! Are you able to check with Segment Support to see if they're sending the same payload multiple times?
I can check, although I would think it's unusual for them to do that... I know they do retries, but those have slightly different timestamps
Based on the behavior seen, that would still be the first thing to check and rule out!
