Data Quality

Data quality concerns whether data is suitable for its intended use. A weekly trend report and an individual payment decision can require different detail, freshness, and evidence. Quality must be assessed against those requirements.

Different checks answer different questions

Completeness asks whether required records and values are present. Uniqueness concerns unwanted duplication. Consistency concerns agreement where facts should agree. Timeliness asks whether data is available when needed. Validity checks permitted formats, ranges, and rules. Accuracy concerns whether values reflect reality.

For example, paid can be an allowed status while being incorrectly assigned to an unpaid order. That value passes a vocabulary check but fails accuracy. A file containing 100 unique IDs may still omit expected orders and contain unrelated ones.

Connect evidence to a response

Compare coverage with an appropriate reference, inspect necessary fields, and record the scope and time of each check. Agreement with a source does not prove the source itself is correct.

Assign an owner and action for failures: investigate, label a result as incomplete, or withhold it where necessary. A single quality percentage is difficult to interpret without its denominator, checks, and remaining limitations.

Reference: UK Government: Data quality dimensions.