Hi Holger,

welcome to the Celonis Community!

What you are searching for is most probably the PU Function. It helps you to aggregate correctly when working on different levels like VBAP and the Activty Table. In the PU Function you can also define conditions (see PQL refernce in the code editor).

For working with the PU Functions, please have a look in the Celonis Help Material and additionally you can find a very good guide here:

The longest but ultimate guide for being a Pull-Up-Functions expert Technical Implementation

During my work with customers I have noticed that there is one specific chapter in PQL where people struggle the most: Pull-Up-Functions (PU functions). The reluctance can range from It works but I dont really understand why to I dont even know what they are. For some just hearing the word Pull-function even creates nightmares . Therefore, I have now taken the time to create a guide that addresses this issue. The following guide on the one hand covers the technical foundation of PU fu

Hope this helps!

Best regards,

Celonis Data Science Team

Hi Holger,

thank you for reaching out.

Actually the built-in KPI works as intended, as we can visualize both the ratio based on cases and based on activities.

Depending on the use cases, boths view might be valuable, as the case-based ratio shows you the number of unique IDs flowing a certain path and the activity-based ratio can show you detailed information about an unwanted activity occurring in XX% of activities.

Please see below an example on how both ratios can be displayed. Choose the option that fits your needs best.

In case of any question, please feel free to reach out.

Best regards,

David Buchmann

Dear David,

works as designed

The correct solution (to be consistent to the process explorer) is provided by the following formula:

COUNT (DISTINCT CASE WHEN _CEL_O2C_ACTIVITIES.ACTIVITY_EN=<%= check_act %> THEN _CEL_O2C_ACTIVITIES."_CASE_KEY" ELSE NULL END)

where the Variable <%= check_act %> contains the desired activity. The result will be the distinct number of cases - dividing them by COUNT_TABLE(CASETABLE) will lead to the equal result as KPI like the process explorer.