Questions on Metric Value Changes for Closed Months

Francesca D.

Hi everyone, can I ask if someone have faced changes about value of the metric calculated for month already closed?

10 comments

Vlad S.
·
·
depending on what you're calculating and how you're calculating it values in the past can change as new data comes in; for example, say you have a retention report set to "on or after", if a user doesn't come back in the first few weeks but then comes back today then your retention value for when the user first interacted will go up... similar things can happen with funnels and conversion windows, etc.
Vlad S.
·
·
can't really tell you any more without knowing more about what kind of report you have and the details of how you have it set up
Francesca D.
·
·
Thank you Vlad. In my case I am counting the event about the session start during a period, in this case it is during a specific month, February/March and so on...So my expectations is that these value can not change. How can I check if some change is coming into and it could be correct or not?
Francesca D.
·
·
Can you support me please about this issue? Can we arrange a call?
Vlad S.
·
·
sorry, I'm not support, I'm just another user 😄
Francesca D.
·
·
Thank you Vlad:)
Vlad S.
·
·
how far back did the values change? like a week back, a month back? also, are you sending events through the ingest API endpoint (which allows you to send events older than 5 days)?
Francesca D.
·
·
I saw the change of the value between yesterday and today... The values are from three month before, and until one month before these values are changing
Vlad S.
·
·
here's some possibilities:
you/someone on your team sent a bunch of events through the ingestion API yesterday, timestamped three months ago
between yesterday and today you changed the session definition settings in mixpanel
you're sending the events through some intermediary (like Segment) and somehow a bunch of events got stuck in the system between then an now, and finally managed to go through on retry
mixpanel was experiencing some data delay
you/someone on your team delete a bunch of events three months in the past through the event deletion request interface
you're counting "unique users" and somehow what you initially thought were distinct anonymous users turned out to be the same person after user merge
Francesca D.
·
·
Very useful the use case you have mentioned, thank you so much. For the second point, we have not done any changes about input value and about the segregation of the environment we are evaluating. About the first point, do we need to analyse logs for understanding the timestamp?or can we use another way? Regarding the third point, how can I get this evidence? For the fourth and the fifth, I need to expect a decreasing values, instead they are increasing The last one, I have already checked and we had the filter by name for example, and also we know that we do not have any case of 2 people with the same name, HR info;)