Data steward and data governance roles: glossaries and ownership
In this article (10 sections)
- Why "customer" needs an owner
- The artefact: a business glossary entry
- Ownership in one table
- A steward's working week
- Owned terms, resolved issues and passed access reviews
- Data owners, the CDO office and compliance
- Confused with data quality analyst, privacy and compliance roles
- Collibra, data catalogue and metadata signals
- Ten terms defined and owned beats a hundred drafted
- Governance lead, data quality manager or CDO office
Related programme: the Data Analytics with Generative AI course (3–4 months; ₹90,000 (inclusive of GST) · EMI from ₹4,000/month).
A data steward looks after a specific domain of an organisation's data, such as customer, product or finance data, keeping its definitions, quality rules and access agreed and documented. Data governance analysts and managers set up the wider framework that stewards work within: policies, the business glossary, the data catalogue, ownership and issue processes. These roles are most common in banks, insurers, telecoms, pharma and large GCCs, where regulators and auditors ask hard questions about data.
Why "customer" needs an owner
In a hypothetical Indian bank, the retail team counts a customer as anyone with an active account, the cards team counts anyone with a card even if dormant, and the regulatory reporting team counts unique customer IDs after de-duplication. Each number is defensible; together they confuse the board. Governance does not pick a winner by decree. It makes each definition explicit, names who owns which, and ensures reports say which one they use.
The artefact: a business glossary entry
This is the basic unit of a steward's work. A good entry is specific enough that two analysts would write the same SQL from it.
Term: Active Retail Customer
Domain: Customer (Retail Banking)
Data owner: Head of Retail Banking Operations
Data steward: [Name], Retail Data Steward
Definition: A unique retail customer (de-duplicated customer ID) holding at least one
savings, current or term deposit account in status ACTIVE on the last
calendar day of the reporting month.
Excludes: Accounts in status DORMANT, FROZEN or CLOSED; customers whose only
relationship is a credit card or loan; staff accounts flagged STAFF.
Source of truth: core_banking.customer_master joined to core_banking.accounts
Calculation: COUNT(DISTINCT customer_id) where account_status = 'ACTIVE'
and product_type in ('SA','CA','TD') and staff_flag = 'N'
Related terms: Card Customer, Total Relationship Customer, Dormant Account
Quality rules: customer_id not null; one customer_id per PAN-verified individual
Used in: Monthly retail MIS; board pack customer count
Version: 2.1, approved 2026-08-14 (changed: staff accounts now excluded)The version line is easy to skip and important. When the board pack number drops one month because staff accounts are now excluded, the glossary explains why before anyone asks.
Ownership in one table
| Activity | Data owner | Data steward | Data engineering | Governance office |
|---|---|---|---|---|
| Approve definition changes | A | R | C | C |
| Maintain glossary entry | I | R, A | C | C |
| Define quality rules | A | R | C | C |
| Implement quality checks in pipelines | I | C | R, A | I |
| Resolve quality issues | A | R | R | I |
| Approve access to sensitive fields | A | R | I | C |
R = responsible, A = accountable, C = consulted, I = informed. The most common failure is having no one in the A column for a domain, so every issue becomes a meeting with no decision.
A steward's working week
Monday might start with the quality dashboard: 1.8% of new customer records this week have a missing PIN code, above the agreed threshold. The steward traces it to one sourcing channel's form and raises it with the channel team. Tuesday brings two access requests for fields containing personal data; the steward checks the stated purpose against policy and routes them to the owner. Midweek there is a glossary workshop with the cards team, who want their own "active customer" definition registered. Thursday is audit preparation: showing lineage from a regulatory report back to source. Friday, the steward updates the issue log and closes items that were fixed.
Owned terms, resolved issues and passed access reviews
Coverage of critical data elements with approved definitions and owners, quality scores against thresholds for those elements, time to resolve quality issues, completion of periodic access reviews, and audit findings related to data. Adoption is the hard one: a glossary nobody uses is shelfware.
Data owners, the CDO office and compliance
Stewards often report to a data governance manager within a chief data office, or sit in the business with a dotted line to governance. They work with data owners, data engineering, analysts, information security, the data protection officer or privacy team, compliance and internal audit.
Confused with data quality analyst, privacy and compliance roles
Data quality analysts focus on measuring and fixing defects; many stewards do this too. Privacy roles, such as those supporting obligations under the Digital Personal Data Protection Act, 2023, focus on lawful processing, consent and rights of individuals; stewards support them but do not replace legal advice. Compliance roles focus on regulatory obligations generally. Data governance ties data definitions, quality and access together.
Collibra, data catalogue and metadata signals
- "Business glossary, data catalogue (Collibra, Alation, Purview, Informatica)" — core governance tooling.
- "Critical data elements, data lineage, BCBS 239" — banking governance for risk data.
- "Data quality rules, DQ dashboards" — quality-focused stewardship.
- "Master data management" — often a separate MDM role with operations work.
- "Consent management, data subject requests" — privacy operations.
Ten terms defined and owned beats a hundred drafted
Pick the ten most-used data elements in your domain and get each a definition, an owner and a quality rule. Then fix one recurring quality issue at its source rather than in a report. Then run your first access review.
Governance lead, data quality manager or CDO office
Senior steward, governance manager, head of data governance, data product manager, or roles in privacy and AI governance; see AI governance and risk roles. The Data Analyst programme builds the SQL and metric-definition skills that make a steward credible with engineers and analysts alike.
Frequently asked questions
Is data governance a technical job?
Partly. Data stewards need to read SQL, understand how systems store data and use catalogue or quality tools, but much of the work is agreeing definitions, owners and rules with business teams. Strong communication and patience matter as much as technical depth.
What is the difference between a data owner and a data steward?
A data owner is usually a senior business leader accountable for a data domain, such as customer or product data, and makes decisions about it. A data steward does the day-to-day work of maintaining definitions, monitoring quality and coordinating fixes on the owner's behalf.
Pankit Kumar has 10 years in Data Science & AI, building and shipping production systems in regulated pharma and clinical environments. He is a freelance trainer at Boston Institute of Analytics, AnalytixLabs and Scaler, and has taught this material to thousands of working professionals.
This article is part of our Data Analytics with Generative AI programme — 3–4 months. The full analyst stack — Excel, SQL, Power BI and Python pipelines — then a generative-AI layer you can prove is right.
Explore Data Analytics with Generative AI