Field notes
When your pipeline runs green but the data is wrong
ADF reports success while rows land truncated, mistyped, or missing. Here's why status-green is not data-correct — and what to validate instead.
- Published
- Reading time
- 1 min read
A Data Factory pipeline returning “Succeeded” tells you the orchestration ran. It tells you nothing about whether the data that landed is correct.
Row counts drift. A source column changes type and a downstream cast silently coerces it. A late-arriving file gets skipped. None of these trip a pipeline failure — they trip a data failure, which surfaces days later in a report nobody trusts.
General Validation runs metadata-driven tests against the source and target of each pipeline and tells you, per run, whether the data actually matches the contract you expected.