Hi @FLMike,
Thank you for your reply.
a) we cannot see your flows
So, here is the flow that consumes all the requests I posted in the table above:

Though there are three Dataverse List Rows actions inside the flow, all three are filtered to retrieve only one result from respective table that matches the value received from the request in the trigger. Those three tables have in total 13 rows (Sessions table 4, Shadow table 5, Device table 4 rows). In addition, the complete solution, including this flow, is in testing phase and no new data is being added, nor current is being removed. There are always 13 rows in total in these three tables, and no other flows, apps, users access these tables and data in any sense.
Since I am the only one who can access flow, I can guarantee no changes were made in those days when requests increased.
b) we dont have the internal Microsoft calculation formulas that they use.
Hm... Don't you think it is strange that these numbers fluctuate almost the double and no data has been added to Dataverse tables?
I agree it is the best idea to call MS support and see if they have any relevant explanation for this.
Thanks once again for your response!