Data Governance Roles and Responsibilities

Data governance roles and responsibilities come down to six jobs: an executive sponsor who funds the program and settles disputes, a data governance council (or committee) that approves standards and cross-domain decisions, data owners who are accountable for a domain, data stewards who do the day-to-day work of definitions and quality, data custodians in IT who run the systems and controls, and working groups that tackle specific problems. Every organization needs all six jobs done. A 20-person company does them with three or four people wearing several hats; a 2,000-person company fills each role separately in every domain. This guide defines each role, gives a RACI table you can copy, and sets out realistic time commitments.

Data governance roles and responsibilities, defined

Executive sponsor

A senior leader, usually the COO, CFO, CIO or a chief data officer where one exists, who owns the data governance program at executive level. The sponsor approves the policy, funds remediation and tooling, chairs or attends the quarterly review, and breaks ties when two data owners disagree. Without a sponsor who controls budget, governance decisions wait indefinitely.

Data governance council

The decision-making body, sometimes called the data governance committee or steering group. It is made up of the data owners, chaired by the sponsor or the governance lead, and it approves standards, cross-domain definitions (what “customer” means when sales and finance disagree), classification levels, exceptions that span domains, and the program’s priorities. It meets on a fixed cadence, usually monthly, and keeps a decision log that doubles as audit evidence.

Data governance lead

The person who runs the program day to day: prepares the council agenda, maintains the policy and standards, tracks the issue log and metrics, administers the catalog, and coaches stewards. In a large organization this is a full-time role or a small governance office; in a small one it is a part of the operations or finance lead’s job. This role coordinates; it does not own the data.

Data owner

The business leader accountable for a data domain. Data ownership means the owner answers for the definitions, quality thresholds, classification and access decisions in that domain, and for the business consequences when they are wrong. Owners are chosen by accountability, not by technical skill: the head of sales operations owns customer data, the controller owns financial data, the head of HR owns employee data. One owner per domain; committees do not own data.

Data steward

The person who does the work of data stewardship on the owner’s behalf: writing and maintaining glossary definitions, specifying and reviewing data quality rules, triaging data issues, keeping catalog entries accurate, and answering “what does this field mean” questions. A data steward is usually a senior analyst or operations specialist who already knows the data well. Business stewards sit in the domain; some organizations also name technical stewards who understand the systems.

Data custodian

The IT or data engineering role responsible for the systems that store, move and secure the data: database administrators, platform engineers, data engineers and application owners. Custodians implement what owners decide: access controls, retention jobs, validation rules, backups, pipeline monitoring and the technical side of the catalog. They do not decide what data means or who may see it.

Working groups

Temporary, cross-functional groups formed to solve a specific problem: reconciling customer definitions across the CRM and ERP, designing the golden record for products, or preparing a domain for an AI project. A working group has a steward as chair, a deadline and a deliverable, then disbands. They keep the council from becoming a place where work is attempted rather than decided.

Everyone else

Every employee who creates, edits or uses data has responsibilities: enter data correctly, follow classification and handling rules, report suspected issues to the steward, and use the glossary definitions in their reports. The policy should say so in one sentence.

RACI table for data governance

RACI assigns each activity one role that is accountable (A), roles that are responsible for doing it (R), roles consulted before (C) and roles informed after (I). Copy this table into your framework and adjust it; the point is that every row has exactly one A.

Activity	Sponsor	Council	Governance lead	Data owner	Data steward	Custodian
Approve the data governance policy	A	R	R	C	I	I
Approve a standard (classification, retention, quality)	I	A	R	C	C	C
Define a business term in the glossary	I	I	I	A	R	C
Resolve a cross-domain definition conflict	C	A	R	R	C	I
Set quality thresholds for critical data elements	I	I	C	A	R	C
Measure and report data quality	I	I	A	I	R	R
Triage and assign a data issue	I	I	C	A	R	C
Fix a defect at the source	I	I	I	A	C	R
Classify a dataset	I	I	C	A	R	I
Approve access to restricted data	I	I	I	A	C	R
Implement access controls and retention jobs	I	I	C	I	C	A
Approve an exception within one domain	I	I	I	A	R	C
Approve an exception across domains	C	A	R	C	I	I
Onboard a new dataset or report	I	I	C	A	R	R
Select governance tooling	A	C	R	C	C	R
Report program metrics to leadership	I	C	A	C	R	I
Fund remediation and tooling	A	C	R	C	I	I

Two rows cause most arguments. “Fix a defect at the source” is accountable to the owner even though the custodian does the technical work, because the owner has to decide the fix is worth the effort and the business process has to change. “Approve access to restricted data” sits with the owner, not IT, because IT cannot judge whether a job needs the data. The data governance framework is where the agreed RACI is recorded.

The small-company version

A 20-person company has the same six jobs and perhaps four people to do them. The data governance operating model is centralized by default, and that is fine. Map the roles to real people like this:

Role	Who does it in a 20-person company	What it looks like
Executive sponsor and council	The founder or managing partner	Approves the two-page policy, settles the rare dispute, joins the monthly review
Governance lead	The operations or finance lead	Runs the thirty-minute monthly review, keeps the glossary and issue log current
Data owners	Three or four department leads	Customer data: head of sales. Financial data: finance lead. Employee data: whoever runs HR. Product or service data: head of delivery
Data stewards	The same owners, or one analyst who knows the systems	Maintains definitions, runs ten quality checks a month, triages issues
Custodian	The IT contractor or the person who administers the CRM and accounting system	Sets up required fields, access rights, backups and exports
Working groups	Two people for two weeks	Example: reconciling client IDs between the CRM and the invoicing tool

The things not to skip, even at this size, are the written owner register (one page), the glossary, and the monthly review on the calendar. They are what make the program survive when one of the four people leaves. Our data governance best practices guide covers the habits that keep the small version running.

A 2,000-person company fills each role separately: a sponsor at executive level, a council of eight to twelve domain owners, a governance lead with a small team, a named steward in every domain, custodians in each platform team, and working groups formed per problem. The model usually becomes hybrid: central standards and tooling, domain-level ownership and stewardship.

Time commitment per role

The numbers below are planning assumptions we use when scoping a program, not measurements. Setup takes more time than steady state; the steady-state figures are what the role costs once the first ninety days are done.

Role	During setup (first 90 days)	Steady state	Notes
Executive sponsor	Two to three hours a month	One to two hours a quarter, plus escalations	Attendance matters more than hours; a sponsor who skips the quarterly review signals that governance is optional
Council member (data owner)	Three to four hours a month, including definition workshops	Two hours a month: one council meeting plus preparation	Owners who cannot give two hours a month are the wrong owners
Governance lead	Half to all of their time, depending on scope	Half a day a week in a small company; a full-time role or team in a large one	Often the first dedicated hire once a program spans more than three domains
Data steward	A day a week per domain	Two to four hours a week per domain	Rises during migrations, audits and AI data preparation
Data custodian	Within normal IT or engineering work, with specific tasks scheduled	Absorbed into platform operations	Budget separately for remediation projects such as adding validation or building quality checks
Working group member	Two to five hours a week for the life of the group	None; groups disband	Set a deadline when the group is formed

If the total looks high, it is because the time is already being spent, invisibly, on reconciling reports, chasing definitions and re-answering questionnaires. Governance moves that time to a schedule and reduces it. If you are new to the discipline, our guide to what data governance is explains where the roles fit among policies, quality and tooling.

Where PS3G fits

PS3G provides data governance consulting and implementation for organizations of every size. Defining the roles, naming the people, agreeing the RACI and training owners and stewards is one of the first deliverables in our 90-day start, because nothing else in governance works until it is done. We facilitate the owner workshops where definitions get agreed, set up the council’s agenda and decision log, and can act as the governance lead under a managed-services agreement while you build the role internally. We also build and run our own regulated software products, so we know what these roles cost in practice as well as on paper.

Related reading

• Data Governance Framework: Components, Examples and a Starter Template
• What Is Data Governance? A Plain-English Guide for Growing Companies
• Data Governance Best Practices: 12 That Actually Stick
• Data governance consulting and implementation services

Frequently asked questions

What is the difference between a data owner and a data steward?

The data owner is accountable for a domain: they approve definitions, set quality thresholds, decide who may access the data and answer for the business consequences when it is wrong. The data steward does the day-to-day work on the owner’s behalf: writing definitions, maintaining quality rules, triaging issues and keeping the catalog accurate. Owners are usually department leaders; stewards are usually senior analysts who know the data.

Who should be on a data governance committee?

The data owners of the in-scope domains, the governance lead, and the executive sponsor as chair or standing attendee. Add a representative from IT or data engineering for custodial questions and, where relevant, from security, privacy or compliance. Keep it to the people who can make decisions; subject matter experts join working groups instead. Eight to twelve members is a practical ceiling.

Does a small company need a data steward?

Yes, but not as a job title. In a 20-person company the data steward is usually the operations lead or an analyst who already knows the CRM and accounting system, spending a few hours a month maintaining a glossary, running quality checks and triaging issues. What matters is that someone is named and the time is protected, not that a position is created.

What does a data governance lead do?

The governance lead runs the program: prepares council agendas, maintains the policy and standards, tracks the issue log and metrics, administers the catalog, coaches stewards and reports to the sponsor. The role coordinates and facilitates but does not own data; ownership stays with business leaders. In larger organizations it grows into a governance office with analysts and a catalog administrator.

Should IT own data governance?

No. IT should hold the custodian role, running systems and implementing the controls that owners decide, and often the governance lead role, because IT sees across domains. Data ownership belongs with the business leaders who understand what the data means and depend on it. A program owned entirely by IT tends to document tables rather than business meaning and struggles to get owners to act.