Skip to main content

Configure: Test

The Test tab is where you validate that your mappings and transformations produce correct results before pushing data to production. Think of it as a quality gate between "I've configured my pipeline" and "I'm confident it works."

test-results

Why Test?​

Mappings can look correct visually but produce unexpected results. Common issues include:

  • Data type mismatches (a string field mapped to an integer column)
  • Join logic that produces duplicate rows
  • Null values where they shouldn't be
  • Incorrect aggregations
  • Missing data from incomplete mappings

The Test tab catches these issues before they affect your downstream analytics.

What You Can Test​

dbt Tests​

These are the data quality tests you configured in the Design tab:

  • not_null — verify that required columns have values
  • unique — verify that key columns don't have duplicates
  • accepted_values — verify that values are within expected ranges

Runtime Tests​

More advanced tests that check:

  • Regex patterns — do values match expected formats (e.g., valid email addresses)?
  • Volume checks — does the table have the expected number of rows?
  • Drift detection — has the schema or data distribution changed since the last run?

Pipeline Execution Test​

Run your entire pipeline (or a subset) in test mode to see:

  • Whether all mappings execute successfully
  • What the output data looks like
  • Any errors or warnings generated during execution

Running Tests​

  1. Select which tests to run (or run all)
  2. Click Run Tests
  3. Review results as they come in

Each test shows:

  • Pass/Fail status
  • Details — what was checked and what the result was
  • Error messages — if a test failed, why it failed

Test Results History​

The Test tab keeps a history of past test runs, so you can:

  • Compare results over time
  • See if a change you made fixed a previous failure
  • Track overall data quality trends

Using the Pipeline Agent​

The Pipeline Agent can help with testing:

  • "Run all tests for the customers table"
  • "Why is the unique test failing on customer_id?"
  • "Add a not_null test on the email column"
  • "Show me the test results from the last run"

Tips​

  • Run tests after every significant change to your mappings or CTEs
  • Start with basic tests (not_null, unique) on your key columns — these catch the most common issues
  • If a test fails, check the Transform tab to review the mapping or CTE that produced the bad data
  • Use volume checks to make sure your pipeline isn't silently dropping rows
  • A playback error (test result differs from a previous successful run) usually means your latest change broke something — investigate before proceeding
Last updated on