Metric views

Metric views is a Databricks data governance capability, introduced May 2025.

Unity Catalog securable objects that define business KPIs once as reusable SQL metrics, separating measure definitions from the fields used to group, filter, and aggregate them.

• They are the core implementation of Unity Catalog semantics, which makes them the rare governance object that dashboards, notebooks, alerts, Genie Agents and external BI tools all read from the same definition. 🔗
• It is a view whose columns you cannot simply select: a measure has no pre-determined level of aggregation until query time, so SQL has to ask for it through a MEASURE function instead. 🔗
• Databricks showed them off at Data + AI Summit 2025 under the name Unity Catalog Metrics and promised general availability 'later this summer'. 🔗

⚠ Limitations: Measures have no fixed aggregation level, so a query must request them through the MEASURE function rather than selecting the column directly. Joined tables cannot include MAP type columns, a metric view that defines parameters cannot be materialized, and a one_to_many join restricts its materializations to exact match with no rollup match. Metric views need Databricks Runtime 16.4 or above, and Microsoft removed BI compatibility mode from the Power BI Databricks connector in April 2026, so Power BI reports that queried metric views through it no longer function. 🔗

Category

Data governance

Introduced

May 2025🔗

Also known as

Unity Catalog metric views, Unity Catalog Metrics, Metric view

Verified

2026-08-20