Dimension built by re-reading the fact query
"Dimension built by re-reading the fact query" is one of the performance checks Auditabi runs on a Power BI semantic model. n table(s) start with `Source = <fact query>` and deduplicate it. It is a medium-severity finding: the model still works, but results are wrong in some filter contexts, slow, or fragile under change. Build dimensions from the master data source (product, customer tables) or from a staged, deduplicated query with Enable load off that the fact and dimensions both reference.
What is detected
n table(s) start with `Source = <fact query>` and deduplicate it. Power Query evaluates the fact query once per referencing table, so the source file or database is read again for every dimension, and Table.Group / Table.Distinct over the full fact is O(rows).
Why it matters
Severity depends on how many objects are affected.
How to fix it
Build dimensions from the master data source (product, customer tables) or from a staged, deduplicated query with Enable load off that the fact and dimensions both reference.
Does your model fail this check?
Free static audit with the exact objects, no account, nothing stored.