What causes SeedColumnTypeMismatch?

A seed CSV column contains mixed values (e.g., integers and strings) that dbt infers as VARCHAR, but the existing seed table in the database still has the old INTEGER type from a previous run. The column_types configuration in dbt_project.yml or the seed YAML specifies a type that is incompatible with the CSV data (e.g., DATE for a column that contains timezone-aware timestamps). The seed CSV was updated with a new column or a changed column format, but the existing seed table was not dropped and re-created with --full-refresh

How do I fix SeedColumnTypeMismatch?

Step 1: Run dbt seed --full-refresh --select to drop and re-create the seed table with the updated schema inferred from the current CSV.. Step 2: If the type is wrong due to inference, add an explicit column_types entry in the seed's YAML: column_types: {id: integer, created_at: timestamp}.. Step 3: Check the CSV for mixed-type columns — a single non-numeric value in an otherwise numeric column will cause dbt to infer the entire column as string.. Step 4: Run dbt seed after applying the fix to confirm the load succeeds.

Medium severitydata_quality

Power BI Refresh Error:
SeedColumnTypeMismatch

What does this error mean?

A dbt seed load failed because the inferred or configured column type for a seed file does not match the existing table column type in the database, causing the COPY or INSERT statement to fail with a type cast error.

Common causes

1A seed CSV column contains mixed values (e.g., integers and strings) that dbt infers as VARCHAR, but the existing seed table in the database still has the old INTEGER type from a previous run
2The column_types configuration in dbt_project.yml or the seed YAML specifies a type that is incompatible with the CSV data (e.g., DATE for a column that contains timezone-aware timestamps)
3The seed CSV was updated with a new column or a changed column format, but the existing seed table was not dropped and re-created with --full-refresh

How to fix it

1Step 1: Run dbt seed --full-refresh --select <seed_name> to drop and re-create the seed table with the updated schema inferred from the current CSV.
2Step 2: If the type is wrong due to inference, add an explicit column_types entry in the seed's YAML: column_types: {id: integer, created_at: timestamp}.
3Step 3: Check the CSV for mixed-type columns — a single non-numeric value in an otherwise numeric column will cause dbt to infer the entire column as string.
4Step 4: Run dbt seed after applying the fix to confirm the load succeeds.

Frequently asked questions

Does dbt automatically run --full-refresh for seeds when the schema changes?

No. dbt seeds do not auto-detect schema changes. You must pass --full-refresh explicitly to drop and re-create the seed table with a new schema.

What is the safest way to change a seed column type?

Update the column_types config in the seed YAML to the desired type, then run dbt seed --full-refresh --select <seed_name>. This ensures the table is re-created cleanly without residual type conflicts.