How do you normalize data for use in Tableau?

Normalising here usually means shaping the data so Tableau can work with it, rather than database normal forms.

The most common operation is pivoting wide data to tall. A spreadsheet with one column per month needs those columns turned into rows of month and value before it can be charted — select the columns in the data source page and choose Pivot, or unpivot in Tableau Prep.

Beyond that: split combined fields, fix dates stored as text, remove duplicate rows, and standardise inconsistent category labels so the same thing is not counted twice.

Structurally, Tableau performs best against a star schema — one fact table surrounded by dimension tables. Where the source is heavily normalised into many small tables, flattening dimensions into wider ones usually helps.

Where the work is repeatable, do it upstream — in the warehouse, or in a Prep flow — rather than rebuilding it in every workbook.