Data steward

What is a data steward?

A data steward is the person who looks after the quality, meaning, and day-to-day usability of a specific area of data. That area might be customer data in the CRM, product data in the webshop, supplier data in the ERP, or finance data in the accounting system.

The role exists because governance does not apply itself. Data governance can say who owns a dataset, what definitions apply, and which quality rules matter. Someone still has to keep those definitions up to date, follow up duplicate records, answer questions, and notice when a field changes meaning.

Stewardship is that practical work. It is less about writing policy and more about keeping data understandable enough for people to use.

What does a data steward do?

The exact job differs by organisation, but four responsibilities return almost everywhere.

• Guard definitions
  Does an inactive customer still count as a customer? Does revenue include VAT? Is a cancelled order part of order volume? The steward helps the business agree on one definition and keeps that definition visible.

• Follow up data quality
  Duplicate customers, missing VAT numbers, invalid product codes, old addresses, and inconsistent categories all need a route to resolution. The steward does not always fix every record personally, but makes sure the issue has an owner and a priority.

• Act as the first point of contact
  When someone does not understand a field, sees a strange dashboard number, or wants to add a new value to a list, the steward knows who should decide and what the current rule says.

• Maintain metadata and documentation
  The steward keeps field descriptions, glossary terms, ownership, known issues, and source notes up to date, whether that lives in Microsoft Purview, another data catalog, or a shared document at the start.

Steward, owner, and IT

Data roles are easy to blur, so it helps to separate decision, operation, and technology.

The data owner is the business decision-maker. Usually this is a manager or domain lead: sales for customer data, finance for financial data, operations for logistics data. The owner decides priorities, access rules, budgets, and final definitions.

The data steward does the operational follow-up. The steward documents, checks, explains, escalates, and keeps the data usable.

IT or the data custodian manages the technical side: systems, backups, integrations, permissions implementation, monitoring, and platform stability.

A simple rule of thumb: the owner decides what should be true, the steward checks whether it is true today, and IT implements the technical setup.

In an SME, the same person may be both owner and steward. That is fine. The distinction matters when decisions and daily follow-up start falling between people.

What does stewardship look like in an SME?

In a large company, data steward may be a formal job title. In an SME, it is usually a hat someone already wears.

The accountant who reviews unpaid invoices and corrects customer records is doing stewardship on finance data. The sales lead who cleans the CRM before a campaign is doing stewardship on customer data. The warehouse colleague who keeps product codes and units consistent is doing stewardship on product data.

The useful step is to make the role explicit. Name one contact per important data domain. Give that person time, a route for reporting issues, and the mandate to ask colleagues to follow the agreed rules. The title matters less than the fact that someone is visibly responsible.

What does a steward need?

Time
Stewardship added on top of a full workload will vanish. Reserve a recurring slot for issue follow-up, documentation, and quality review.

Mandate
The steward must be allowed to challenge sloppy entry, reject unclear changes, and escalate bigger decisions to the owner without making every small correction a committee meeting.

Tooling
Automated quality rules, a data catalog, lineage, and a glossary reduce manual work. Microsoft Purview Unified Catalog, for example, uses governance domains and roles such as Data Steward and Data Product Owner. But a first step can still be a shared list of definitions, owners, and known issues.

Management backing
Data quality often conflicts with speed. A salesperson wants to create a customer record quickly. The steward wants the VAT number and country code filled in correctly. Management has to support the rule, or the steward loses the argument every time.

What to watch out for

A role without time is just a name
Putting a person in a spreadsheet as steward costs nothing and changes nothing. Plan the work like other operational work.

One steward for everything rarely works
Customer data, product data, and finance data need different business knowledge. Assign stewards close to the data they already use.

Do not make stewardship an IT dumping ground
IT can build a validation rule. It cannot decide which customer address is the correct one when sales and finance disagree.

Capture knowledge before people leave
If the steward leaves and all definitions lived in that person's head, the organisation is back where it started. Keep definitions and decisions in a shared place.

Keep the loop short
A steward who only reports issues but cannot get fixes scheduled will burn out. Make sure quality problems flow into real backlog, cleanup, or source-system changes.