Why dimensions matter
“The data looks wrong” is difficult to manage. Data quality dimensions provide a structured vocabulary for defining rules, measuring problems, and assigning ownership.
Seven core dimensions
| Dimension | Question |
|---|---|
| Accuracy | Does the value correctly represent the real-world fact or authoritative source? |
| Completeness | Is all required data present? |
| Consistency | Does the same concept agree across datasets or systems? |
| Validity | Does the value conform to format, range, type, and business rules? |
| Uniqueness | Are duplicate records appropriately controlled? |
| Timeliness | Is data available and updated when needed? |
| Integrity | Are keys and relationships preserved correctly? |
Accuracy
Completeness
Completeness may mean required fields are non-null, required records arrived, or an expected period contains all data.
Consistency
Consistency failures appear when systems disagree—for example, an account marked Active in CRM and Closed in the warehouse.
Validity
Validity checks include type, format, allowed values, ranges, cross-field rules, and business constraints.
Uniqueness and duplicates
Not every repeated value is a duplicate. Define the business key and duplicate rule explicitly.
Timeliness and integrity
Timeliness is critical for operational or near-real-time decisions. Integrity ensures parent-child and reference relationships remain valid after processing.