Ad hoc, spreadsheet-style formulas applied to a query's results after it runs — shown as green columns in the data table.
Require the create_table_calculations permission and can only reference fields already in your query.
Common shortcuts: % of column, % change from previous row, running total, rank.
Limitations: formulas must be lowercase, decimals under 1 need a leading zero, calculations only cover rows actually returned, and any calc using offset() can't be sorted.
Create new dimensions and measures in an Explore without touching LookML — requires the create_custom_fields permission.
Unlike table calculations, custom fields generate real SQL and run against the database.
A custom measure can be built from a dimension (sum, average, count distinct, min, max, median), or as a filtered custom measure layered on an existing measure.
A custom dimension can be built via a Looker expression, including custom groups and custom bins for bucketing values.
Logical comparisons and booleans, plus Positional functions like offset and pivot_index for pulling values from other rows or pivots.
Functions that aggregate across rows, convert data types, or reference other rows/pivots only work in table calculations — not in custom filters or custom fields.