1. Accuracy: Does the value reflect reality?
Accuracy asks whether the stored value correctly represents the real-world entity or event. A contact title can be formatted perfectly and still be inaccurate if the person changed roles. A company website can be syntactically valid and still belong to the wrong organization.
Accuracy usually requires comparison with a trusted source. For high-impact fields, keep source context or a verification date so the team can understand why the value is trusted.
2. Completeness: Are the required values present?
Completeness measures whether the fields needed for the intended workflow are populated. It should be defined against required fields rather than against every possible column.
A prospecting file may be complete when Company Name, Website, First Name, Last Name, Title, Direct Email, and LinkedIn URL are present. A different project may require industry, location, employee range, or custom research fields.
3. Consistency: Do equivalent values follow the same rules?
Consistency asks whether data is represented uniformly across records and systems. Problems appear when one dataset uses state abbreviations while another uses full names, when industry labels use several taxonomies, or when the same company has different canonical names in different tools.
Normalization rules improve consistency. Decide which format is authoritative before merging or reporting rather than standardizing differently each time.
4. Timeliness: Is the data current enough for the decision?
Timeliness is contextual. A legal company name may remain useful for years, while a contact title or technology signal can become outdated quickly. The question is not whether the data is new, but whether it is current enough for the task.
Use verification dates and refresh rules for fields that change frequently. Prioritize refresh work where stale values create real cost, such as outreach, routing, compliance, or migration.
5. Validity: Does the value conform to the required rule?
Validity measures whether a value follows the allowed format, type, range, or rule. Examples include a date using the expected format, a state field using an approved value, or an employee count being numeric and nonnegative.
Validity is often easier to automate than accuracy. A value can be valid without being true, so validity checks should not be treated as proof of accuracy.
6. Uniqueness: Are duplicate records controlled?
Uniqueness evaluates whether an entity or identifier appears more than intended. Duplicate CRM accounts, repeated contacts, duplicated product records, and repeated company domains can distort counts and trigger duplicate work.
Uniqueness rules need a matching strategy. Exact duplicates are easy. Near-duplicates involving abbreviations, subsidiaries, alternate domains, or name variations may require entity resolution and human judgment.
Use the six dimensions as a scorecard, not a checklist
The six dimensions become useful when they are tied to measurable rules. A team might track required-field completeness, duplicate rate, percentage of records verified within a target time window, or validity failures by field.
Do not weight every dimension equally by default. A dataset used for outreach may place more weight on accuracy and timeliness. A migration project may prioritize consistency, validity, and uniqueness. The intended use should determine the quality threshold.
- Accuracy: correct compared with reality
- Completeness: required values are present
- Consistency: equivalent data follows the same conventions
- Timeliness: current enough for the intended use
- Validity: conforms to defined rules
- Uniqueness: unwanted duplicates are controlled
Common questions
What are the six dimensions of data quality?
A widely used set is accuracy, completeness, consistency, timeliness, validity, and uniqueness.
Are there only six data quality dimensions?
No. Frameworks vary. IBM notes there is no universal standard, although these six are widely adopted and provide a practical core for measurement.
What is the difference between accuracy and validity?
Validity asks whether a value follows the defined rule or format. Accuracy asks whether the value reflects reality. A properly formatted value can be valid but still inaccurate.
Which data quality dimension matters most?
It depends on the workflow. The most important dimensions are the ones tied to the decisions, automations, reporting, or operational risks the dataset supports.
Final thoughts
Data quality is not one score. A dataset can be complete but inaccurate, validly formatted but stale, or current but full of duplicates. A widely used framework evaluates six dimensions: accuracy, completeness, consistency, timeliness, validity, and uniqueness. IBM notes that there is no universal standard for every data-quality framework, but these six remain widely adopted in practice.
Need help improving your business data?
DataDabble can research, clean, enrich, and maintain the records your team depends on.
