## TL;DR
Change the column type through a replace, not a merge: Delta MERGE refuses to change an existing column type, so rewrite the table with the new type in a controlled replace, or land the widened values in a new column and migrate readers. Delta Lake protects existing column types during merge. A safe widening such as int to long still fails inside MERGE, because merge never rewrites the schema of matched files.

```text
delta lake merge failed error "cannot change data type of existing column" on schema evolution
```

## Use this when
- The check or run fails with this symptom: a Delta Lake merge fails with cannot change data type of existing column.
- You are working on a Delta table updated by MERGE whose source column type changed and the failure repeats on the same data or configuration.
- You need the failing rows, rule, or stage isolated before changing thresholds or code.

## Not for this skill when
- A narrowing change such as string to int, which needs value validation and cleansing rather than a schema rewrite.
- The error text names a different tool, stage, or root cause. Match the verbatim error first and use that skill.

## Steps
1. Read the merge error and identify the column and the from and to types. Verify: the exact column and type pair are known
2. Check whether the change is a safe widening, int to long or float to double, or a real conversion. Verify: the change class decides the fix path
3. For a widening, rewrite the table once with the new schema outside the merge, using a replace or a create-or-replace from a typed select. Verify: the table schema shows the new type and row counts match before and after
4. For a conversion, add a new typed column, backfill it with a cast that surfaces bad values, and switch readers before dropping the old column. Verify: bad values are listed instead of silently nulled
5. Rerun the original merge. Verify: the merge completes with the schema already at the target type

## Variant phrasings
### delta lake merge cannot change data type of existing column
The shorter phrasing focused on the named failure.
### delta merge schema evolution type change failed
The symptom phrasing a searcher tries next.
### cannot change data type existing column delta fix
The variant naming the failing stage or tool directly.

Compatibility: Delta Lake 2.x and 3.x on Spark and Delta standalone writers

## Why it happens
MERGE updates files in place and validates the source schema against the target schema strictly. Type changes are a schema evolution operation that must happen as a table rewrite, so a pipeline that only ever merges has no path to apply them.

## Edge cases / pitfalls
- Enabling schema merge options on MERGE does not permit type changes; it only allows adding columns.
- A replace rewrite drops table history beyond the retention used for the rewrite; snapshot the table first if time travel matters.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_pUXxZ-IlPRmyGBiUZdVLqg