@szymon_dybczak thanks for the reply. I have a single job that processes tables of multiple domains. I cannot split else the costs will blow up. The granular cost monitoring link shared talks about costs attribution for sql warehouses. We already looked at the math being applied there - compilation time + execution time + xxxxx is considered to attribute costs to users.

In our case, it is a job that runs on job compute. I hope I clarified the situation.