Data Tests Overview

Data tests are useful for validating and enforcing explicit expectations on your data.

Elementary enables data validation and result tracking by leveraging dbt tests and dbt packages such as dbt-utils, dbt-expectations, and Elementary. This rich ecosystem of tests covers various use cases, and is widely adopted as a standard for data validations. Any custom dbt generic or singular test you develop will also be included.
Additionally, users can create custom SQL tests in Elementary, and add tests in bulk from the UI.

The combination of dbt tests, Elementary monitors, custom SQL tests and the rich dbt testing ecosystem provides the ability to achieve wide and comprehensive coverage.

​

Supported data tests

​

Test configuration

One of the design principals in Elementary is that users should manage configuration in code. This enables to maintain the same workflow for building the pipeline and configuring coverage, it makes observability and governance part of the development cycle, and provides control in review process and version management.

However, adding many tests in code is tedious, and configuration in code isn’t usable for everyone.

In Elementary, we designed a flow to incorporate the good of both worlds:

• Configuration in code or in UI - The UI test configuration flow opens pull requests to the code base through the code repository integration.
• The code is the single source of truth - As configuration from UI goes to code, the code remains the place where configuration is managed and maintained.

​

Create new tests from the UI

​

Edit existing tests from the UI

Tests can be edited from the UI, and the changes will create a pull request in the code repository. The pull request will be reviewed and merged by the team, and the changes will be applied to the tests after dbt pipeline is run and Elementary is synced. Simply open the test side panel from the test results / incidents page, and navigate to the Configuration tab. Then Click “Edit test” and make the necessary changes.

A pull request will be opened in the code repository, and a link to the PR will be provided in the UI.