Data Quality and Governance
How to judge whether data is fit for use and establish clear organizational responsibility for managing it.
Assessing
| Dimension | What it asks |
|---|---|
| Accuracy | Do values correctly represent real-world facts? |
| Completeness | Are required records and fields present? |
| Consistency | Do related systems and records agree? |
| Timeliness | Is information current enough for its process? |
| Validity | Does it follow defined formats, ranges, and business rules? |
| Uniqueness | Are unintended duplicates avoided? |
Assessing
is not a universal score detached from purpose. Data is fit for use when it is sufficiently accurate, complete, consistent, timely, valid, and unique for the task at hand. Which dimensions matter most depends on the task: a missing customer phone number may hinder a sales campaign, while a missing supplier tax identifier may block a financial transaction.
The six dimensions in the table help turn a broad question—“Can we rely on this information?”—into more specific questions. They are related but not interchangeable. For example, a value might follow an approved format (validity) yet fail to represent the real-world fact correctly (accuracy). Likewise, a record could be accurate when entered but no longer current enough for the process using it (timeliness).
Organizations make expectations measurable by setting checks for particular processes. In an order-to-cash process, for instance, each order might be required to refer to a valid customer and product, and the customer’s billing details might need to meet completeness requirements. Teams can then monitor error rates, missing-field percentages, duplicate counts, and how long changes take to appear in connected systems. These measures make quality expectations observable; the relevant checks should reflect the process rather than assume every field has equal importance.
Quality problems can affect more than one record or system. Conflicting, incomplete, or outdated information may spread through integrated business processes—for example, a duplicate customer record can lead to multiple invoices, while an incorrect product record can affect purchasing, inventory, and sales. Fixing an error at its source is usually more effective than repeatedly correcting downstream copies, because it addresses the originating problem instead of allowing it to recur in connected records.
Establishing
| Role | Responsibility |
|---|---|
| Business leaders | Set priorities and acceptable risk. |
| Data owners | Make decisions about a data domain. |
| Data stewards | Define and monitor rules; help resolve quality issues. |
| Technical teams | Implement controls and integrations. |
Establishing
formally establishes how decisions about data assets are made: it sets decision-making authority, policies, standards, and accountability. It is not solely an IT responsibility. The responsibilities in the table show why governance involves both business and technical work: business leaders set priorities and acceptable risk, while technical teams implement controls and integrations. Data owners and stewards provide domain-level decisions and oversight.
Governance also sets practical rules for managing information. These can specify standard meanings, formats, identifiers, and allowed values; which system or role is authoritative for each data element; who may create, approve, change, share, retain, or delete data; how quality issues are measured and addressed; and how definitions, lineage, and transformations are documented. Assigning these responsibilities makes it clearer who can decide, who maintains the rules, and who puts controls into operation.
For example, an organization might designate its ERP system as authoritative for product cost and inventory quantity, while its CRM is authoritative for sales interactions. This does not mean one system is authoritative for every kind of data. Rather, the designation depends on the data element. A shared customer identifier and documented synchronization rules help connected systems exchange information without confusing a local copy with the authoritative record.
Governance and are connected, but they are not the same thing. Governance establishes standards, decision rights, and accountability; quality describes how well information meets the needs of its intended use. Clear governance can make quality expectations and issue-handling responsibilities explicit, while quality checks help reveal whether those expectations are being met.
: The degree to which data is suitable for a particular task.
: A person responsible for decisions about a data domain, such as customer or product data.
: The system or role designated as the trusted source for a particular data element.