Definition
The practice of checking that data — in databases, APIs, or pipelines — meets expectations for correctness, completeness, and consistency before or after it is used. Data validation includes schema validation (does the data have the right columns and types?), value validation (are values within expected ranges?), and referential integrity checks (do foreign key relationships hold?).
Why it matters
Data validation is the difference between 'we think the data is correct' and 'we know the data is correct.' Without it, bugs reach production silently. With it, bad data is caught before customers feel it. The key question is when validation runs: at pipeline load (Great Expectations), in production monitoring (Soda), or at release time (Well Tested).
How Well Tested handles it
Well Tested performs data validation at release time — checking table diffs, keyed unmatched rows, and aggregate values against production state before a release ships. This is different from continuous monitoring (Soda) or pipeline testing (Great Expectations). Well Tested's validation is specifically about 'is this release going to break something in production?' — not 'is our warehouse data healthy in general?'
Related terms
No commitment. 20 minutes. We review your release workflow and show you what's possible.