What blending actually does
Blending combines data from different sources — an Excel file with an Oracle table, say — without any special coding.
The mechanics matter: Tableau queries each source separately, aggregates each to the level of the view, and then combines the aggregated results on a shared dimension called the linking field. It is not a row-level join. It is an aggregate-level merge that behaves like a left outer join from the primary source.
Primary and secondary
The first data source you use in a sheet becomes primary; it gets a blue checkmark. Others become secondary, with an orange checkmark. Linking fields show a link icon you can click to activate or deactivate.
Because the blend behaves like a left join from the primary, switching which source is primary changes your results. Rows in the secondary with no match in the primary simply don't appear.
Limitations you will hit
Secondary source measures are always aggregated — you cannot get row-level detail from a secondary source. You also can't publish a blended relationship as a single data source, and secondary dimensions used in the view can filter awkwardly.
Blending is also slower than a join when both tables live in the same database, because it means two round trips instead of one.
Joining vs. blending
The rule of thumb: if the data is in one place, join it. Reach for blending when the sources genuinely can't be joined — different databases without cross-database support, or different levels of granularity where a join would fan out badly.
That second case is underrated. If you have daily sales and monthly targets, joining produces duplication; blending aggregates each side independently and lines them up cleanly.