What causes NON_LAST_MATCHED_CLAUSE_OMIT_RESULT?

MERGE INTO written with a WHEN MATCHED clause that has no THEN action, placed before another MATCHED clause. Migrating MERGE syntax from SQL Server or Oracle where the ordering rules differ. A conditional MATCHED clause intended to 'skip' rows being placed in a non-terminal position. Auto-generated MERGE SQL from an ORM or ETL framework producing clauses in the wrong order

How do I fix NON_LAST_MATCHED_CLAUSE_OMIT_RESULT?

Move the WHEN MATCHED clause that omits an explicit action to be the last MATCHED clause in the MERGE statement. Add an explicit `THEN DO NOTHING` (or equivalent no-op) to intermediate MATCHED clauses if your intent is to skip certain rows. Review the MERGE clause ordering: WHEN MATCHED clauses are evaluated top-to-bottom and only the first matching clause fires per row. If skipping rows is the intent, use a negated condition on the preceding MATCHED clause instead

Medium severityquery

Power BI Refresh Error:
NON_LAST_MATCHED_CLAUSE_OMIT_RESULT

What does this error mean?

A MERGE INTO statement contains a WHEN MATCHED clause that does not produce an action (INSERT/UPDATE/DELETE) and is not the last MATCHED clause. Databricks requires that only the final MATCHED clause can omit an action.

Common causes

1MERGE INTO written with a WHEN MATCHED clause that has no THEN action, placed before another MATCHED clause
2Migrating MERGE syntax from SQL Server or Oracle where the ordering rules differ
3A conditional MATCHED clause intended to 'skip' rows being placed in a non-terminal position
4Auto-generated MERGE SQL from an ORM or ETL framework producing clauses in the wrong order

How to fix it

1Move the WHEN MATCHED clause that omits an explicit action to be the last MATCHED clause in the MERGE statement
2Add an explicit `THEN DO NOTHING` (or equivalent no-op) to intermediate MATCHED clauses if your intent is to skip certain rows
3Review the MERGE clause ordering: WHEN MATCHED clauses are evaluated top-to-bottom and only the first matching clause fires per row
4If skipping rows is the intent, use a negated condition on the preceding MATCHED clause instead

Frequently asked questions

Can I have multiple WHEN MATCHED clauses in Databricks MERGE?

Yes — Databricks supports multiple WHEN MATCHED clauses, each with different conditions. They are evaluated in order and only the first matching clause fires per row. The clause without an action must be the last one.