Fitting tests into the development workflow

dbt testing verifies assumptions about your data and models throughout the development workflow, not just at the end. Tests should be run incrementally as you build models, and strategically before merging code or deploying to production to catch issues early.

1 · Learn the must-know

• 1Run dbt build instead of separate dbt run + dbt test commands to test each model immediately after it's created, stopping downstream dependents from running on failure by default.
• 2Use dbt test --select model_name or dbt build --select model_name+ to test a specific model and its downstream dependencies during development.
• 3Generic tests (unique, not_null, accepted_values, relationships) are defined in YAML under a model's columns: or tests: key, while singular tests are standalone SQL files in the tests/ directory.
• 4Tests can be configured with severity: warn versus the default error so that non-critical data quality issues surface without blocking the pipeline.
• 5In CI/CD workflows, dbt tests typically run against a temporary schema built from modified code (often via Slim CI with state:modified+) so only changed and downstream models are tested, saving compute.
• 6The --store-failures config (or store_failures: true) persists failing test records to a database table, making it easier to debug why a test failed without rerunning it.

2 · Check your understanding