Validation asks whether data behaves as expected
It checks schema, required fields, ranges, relationships, uniqueness, formats, and business rules.
Resource guide / Data validation
Turn data-quality expectations into explicit tests
Validation checks whether information is complete, plausible, consistent, and fit for a defined use. It does not prove that every value is true.
Working principle: A passing rule means the record met a stated condition. It is not a substitute for source verification or domain judgment.
Scope
Validation and verification answer different questions
Verification asks whether a claim is supported
It connects the value or assertion to credible evidence, current sources, and an appropriate review decision.
Validation workflow
Build checks around the decision the data will support
Define the use and tolerance
Identify who will use the result, what could go wrong, and which failures should block release.
Write observable rules
Express expectations as testable conditions with clear pass, warning, and fail states.
Test across levels
Check fields, records, relationships, files, time series, and cross-source consistency.
Triage exceptions
Separate actual defects from legitimate edge cases, then assign an owner and resolution state.
Retest and record
Run the same checks after correction and preserve the rule version, result, and reviewer decision.
Rule families
Use a compact set of explainable checks
Completeness
Required fields, expected coverage, missing values, and gaps across time or entities.
Conformance
Types, formats, allowed values, units, and schema compatibility.
Consistency
Cross-field logic, related-table integrity, stable definitions, and agreement across sources.
Plausibility
Ranges, distributions, changes, and domain constraints that surface suspicious values.
Uniqueness
Duplicate identifiers, repeated events, and records that may represent the same entity.
Freshness
Observation dates, update cadence, expired records, and time-sensitive source requirements.
A practical next step
Bring a real question. Leave with a clearer evidence path.
Show us the dataset, claim, report, or market question your team needs to trust. We will map the relevant workflow and the review points it requires.