Data governance glossary
Data governance has a vocabulary problem: the same word means three different things in three different meetings. So here is every term you will actually run into, explained the way you would explain it to a colleague — what it means, how it works on a normal Tuesday, and where it usually goes wrong.
Foundations
The words everything else leans on: what governance is, how far it reaches, and the policies and standards holding it up.
6 terms- DAMA-DMBOK
- DAMA International's reference framework for data management: eleven knowledge areas arranged around data governance at the center.
- Data governance
- The system of decision rights and accountabilities for data — who may decide what about which data, and on what basis.
- Data governance committee
- The executive forum where data owners sit alongside C-level sponsors, the DPO and the security lead to give governance its mandate, its budget and its risk decisions.
- Data management
- The work of building and running the systems, processes, and controls that hold data through its whole life.
- Data policy
- A short, binding statement of what the organization requires of its data, approved by someone with the authority to require it.
- Data standard
- The specific, testable rule that says how a policy is satisfied — format, allowed values, naming, tolerance.
Roles and accountability
Who decides, who answers for it, and who does the daily work of keeping data usable.
5 terms- Data governance council
- The standing forum of data owners that settles the decisions no single domain can make alone.
- Data owner
- The single accountable person for a data domain — the one who decides, approves and answers for it, and who has the budget to act.
- Data steward
- The person who does the day-to-day governance work in a domain: definitions, rules, quality issues, and the questions nobody else can answer.
- Data stewardship
- The practice and the network of people through which governance decisions actually reach the data.
- Decision rights
- The explicit record of which role decides, which is consulted, and which is merely informed — per decision, not in general.
Metadata and catalog
How your organization knows what data it has, what it means, and where it came from.
6 terms- Active metadata ingestion
- Capturing metadata as events happen — in real time, from the systems themselves — and feeding it back into the tools where people work.
- Business glossary
- The agreed, owned definitions of the terms the organization reports on — one meaning per term, with a name against it.
- Data catalog
- The searchable inventory of data assets, with their meaning, owner, lineage and quality attached.
- Data lineage
- The traceable path a value takes from where it originated to where it is used, including every transformation on the way.
- Master data management
- The discipline of maintaining one authoritative record for the entities the whole business shares — customer, product, supplier, employee.
- Metadata
- The descriptive layer around a data asset: what it means, where it came from, who owns it, how it may be used.
Data quality
Whether the data can be trusted, how you measure that, and what to do when the answer is no.
7 terms- Cost of poor data quality
- The measurable money and time an organization loses to data that is wrong, missing, late or duplicated.
- Critical data element
- A data field whose failure has a material consequence — regulatory, financial, operational or reputational — and which therefore gets governed first.
- Data profiling
- Examining actual data to learn what is really in it — value distributions, null rates, formats, outliers, relationships.
- Data quality
- The degree to which data is fit for the purpose someone is using it for — which means quality is always relative to a use.
- Data quality dimensions
- The standard axes along which data can fail: completeness, accuracy, consistency, timeliness, validity, uniqueness.
- Data quality rule
- An executable test on data with a defined threshold and a named owner who acts when it fails.
- DMAIC
- The Six Sigma improvement cycle — Define, Measure, Analyze, Improve, Control — used to fix a data quality problem and keep it fixed.
Literacy and culture
The human side: the skills and habits that keep governance alive after the project ends.
3 terms- Data culture
- The set of habits that determine whether evidence changes what an organization does — including when it contradicts the person in charge.
- Data literacy
- The ability to read, question, interpret and argue with data well enough to make a decision with it.
- Data-driven decision making
- Making choices where the evidence is examined before the conclusion is formed, and can change it.
Maturity and measurement
Where you stand today, where you are heading, and how you show that you moved.
6 terms- Data governance operating model
- The written arrangement of who governs what, how decisions are made, and how the work reaches the data day to day.
- Data maturity
- How reliably and repeatably an organization manages its data — whether the outcome depends on the process or on the individual.
- Data maturity model
- A structured scale for rating data capabilities, usually from ad hoc to optimized, used to diagnose where to invest.
- Levels of maturity
- The named stages a maturity model scores against — typically ad hoc, repeatable, defined, managed and optimized — describing behaviour rather than tooling.
- Proactive data governance
- Governance built into how data is created and changed, so problems are prevented at the source instead of found downstream.
- Reactive data governance
- Governance that only moves when something breaks — an audit finding, a wrong number in a board pack, a breach — and goes quiet in between.
AI, privacy and risk
What has to be protected, from whom, and what AI changes about both questions.
7 terms- AI governance
- The controls over how models are built, approved, monitored and retired — most of which are data controls wearing a new name.
- Confidential data
- Data whose exposure harms the organization — pricing, margins, salaries, deal pipelines, source code, customer lists.
- Data classification
- Labeling data by how sensitive it is, so that handling rules can be applied automatically rather than remembered.
- Dynamic data masking
- Hiding or transforming sensitive values at query time based on who is asking, while the stored data itself stays untouched.
- Personally identifiable information
- Data that identifies a living person, directly or in combination with other data the holder can reach.
- Role-based access control
- Granting access to roles that describe a job, then putting people into roles — so permissions are reviewed and revoked as a set, not one by one.
- Sensitive data
- Data whose exposure harms the person it describes — health, biometrics, beliefs, sexual orientation, criminal records, precise location.
Architecture and data products
How data becomes products, domains and contracts that other teams can build on.
7 terms- Data contract
- An explicit, versioned agreement between a data producer and its consumers covering schema, meaning, freshness, quality and how change is announced.
- Data domain
- A bounded subject area of data — customer, product, employee, contract — that one accountable owner can reasonably answer for.
- Data product
- A curated, documented, owned dataset built and maintained for known consumers, with a service level attached.
- Data subdomain
- A slice inside a data domain with its own steward, its own vocabulary and its own critical data elements.
- External data marketplace
- Where data products are exchanged with parties outside the company — as a buyer of third-party data, or as a publisher of your own.
- Internal data marketplace
- The place inside the company where teams browse published data products, request access, and get it — with an owner, a contract and a service level attached to each listing.
- Single source of truth
- The one place designated as authoritative for a given piece of data, which everything else derives from or defers to.