Ok I found this article which explain is pretty well: 

Although I’m not sure how this is computed. Is it counting active time when there is a continuous stream of events Fired associated with one user ID?

Our results seems way lower of what we had measured before on Google Analytics. Any idea what could be the reason?
 

I’m also not sure if this is measuring time per session on per day. The former would make more sense to me given the low results.

Hey @Moritz ,
I didn’t quite understand the “retention chart” part of your question, but looks like you came across the right chart to measure the time spend by a user in your product i.e the User Sessions chart.

As per the doc, the Time Spent per User is the average amount of time spent in sessions per interval which is calculated by taking the sum of all session lengths within the interval, and dividing by the total number of active users in the interval. The session length for a user_id is calculated as the difference between the session_id set by your app during the start of a session and the timestamp of the last event before that particular session ends.

W.r.t. your data in Google Analytics, there are bound to be differences in the results. One of the common reasons is the way the session timeout window is defined in both of your instrumentation approaches. Here is a good resource to understand this. I would also point out to this post for further reference.

Hope this helps!