Please can someone explain what happens under the hood when we slice beast modes (BMs) by dataset fields and/or when a filter is applied to BMs (categorical field filters/date filters)
BMs using SUM() or COUNT( DISTINCT() ) when sliced or filtered don't behave as expected. (expectations being the issue). In one case, I had duplicates so used COUNT(DISTINCT) and sliced this BM with a categorical field. The field counteracted the distinct function and effectively did a count on the field's values.
The broader issue here is reaching for the right tool. From a DAX reference point, it is easy to forget that calculations usually required a combination of ETL and BM to get the job done.
Hopefully, the responses to this fill in the gaps as to why.