dbt tests for data quality

dbt tests are SQL assertions that run against models and sources. They do not transform data; they return failing rows. Zero failing rows = pass.

Built-in (generic) tests

Usually declared in YAML next to columns:

models:
  - name: fct_orders
    columns:
      - name: order_id
        tests:
          - unique
          - not_null
      - name: customer_id
        tests:
          - relationships:
              to: ref('dim_customers')
              field: customer_id
      - name: amount
        tests:
          - dbt_utils.accepted_range:
              min_value: 0

Custom / singular tests

A SQL file under tests/ that should return 0 rows, e.g. "no shipped order without a ship date."

Commands

• dbt test: run tests only
• dbt build: run models and tests in DAG order so a failed upstream test can block dependents

dbt run   -> build relations
dbt test  -> assert expectations
dbt build -> run + test together (quality gate in the DAG)

Source freshness

dbt source freshness checks whether raw sources updated within warn_after / error_after windows.

Interview tip: Say dbt tests are cheap warehouse SQL checks co-located with models. List unique, not_null, relationships, plus custom SQL and source freshness.