What is data blending?

Blending brings together data from separate sources by aggregating each to the level of the view and matching the results on a shared dimension.

It is not a join. Each source is queried independently, so the two never meet at row level. The combination behaves like a left outer join from whichever source was used first in the sheet.

That design is what makes it good at mismatched granularity — daily actuals against monthly targets, where a join would repeat the target across every day.

The limitations follow from the same design: secondary measures are always aggregated, there is no row-level access to the secondary, and a blended relationship cannot be published as a single reusable data source.