Data Quality Overview

There are many great existing data quality tools out there, primarily focused on runtime validations. There are tools for observability, expectation testing, and more. SDF provides 4 concepts for data quality.

Checks (static) - Are SQL queries run against SDF’s information schema to validate absence of a condition. Checks are executed against compile-time metadata so they provide immediate feedback when developing in the workspace and are typically fast and easy to integrate into CI/CD. They help you guarantee business logic among queries and increase development speed at the same time.
Reports (static) - Are SQL queries run against SDF’s information schema to understand the presence of a condition. Reports are also executed against compile-time metadata and are therefore excellent to gain immediate insights on the entire system, for example for internal or external audits, governance management, and system health tracking.
Tests (dynamic) - Are SQL queries run against your data to validate a certain condition in the data itself, meaning after the statement has been executed. This is helpful for ensuring that the shape of a particular column is as expected, that thresholding conditions are met, or that there are no null values.
Stats (dynamic) - Are SQL queries run against your dataset for at-a-glance understanding. Gain a birds eye view of a particular column including distributions, uniqueness, or other summary statistics configurable by you.

See the diagram below for the order of operations of data quality steps. It is recommended to run static data quality validations like compiling, checking, and reporting as part of CI/CD.