Time between user actions


Hi all,

Within an analysis I would like to show/analyze how much time, from the user perspective, there was between his actions. So instead of within a case saying there was 1 hour between A and B, I would like to know how much time there was for user X between performing A (for case 1) and A (for case 2) for example.

I’m however uncertain how to ‘flip’ the case notion like this without going to the data model and making the user the case (which would allow for all kinds of interesting analyses I suppose).

I wonder if/how this is possible in 4.4.



I do not know an easy way to do it :frowning_face:

There are several workarounds:

  1. Play around social network analysis. With some creative filtering on tab level you can get number of actions “A” per user and see how it changes in time
  2. Create OLAP table with username as dimension and calculate number of unique activities “A”: COUNT(DISTINCT CASE WHEN “activity” = ‘A’ THEN “activity_id” ELSE NULL END) and divide them by time period
  3. Create second data model with user_id + day (month/week) of activity as case key. In this case you will be able to use all standard Celonis tools, but will pay for extra license


Hi Joos,

So this is definitely an interesting use case.

I guess the correct answer to this question would be that it depends on the source of your user information and the link to case and activity table. For example in a P2P process the user information might be linked on header level to the case and activities so this would not work.

What you would need is the information on case or activity level (either because it is in the columns of case and activity table or because you have it in a linked table with a 1:1 relationship). In that case, you could use the user as a dimension and use the source/target functions on the activity column in combination with a days_between function or similar to retrieve the information you are interested in.



Hi Henry,

If I set the user as a dimension, would source/target refer to previous/next actions of that user then? Or next/previous activities for a case?

Currently I’m working with a separate data model (option 3) which gives nice insights, but is of course not scalable.


Hi Joos,

the informationen would be aggregated on user level and thus the KPI would be calculated on user level as well, aggregating all the cases for this user and calculating the KPI. I can not guarantee that this will work as the use case is highly experimental but it could be worth a try.