Know who owns a dataset

A practical reference for data stewards and people who use organizational data.

At fictional Northstar Works, each dataset has a business owner accountable for its intended meaning and use, and a data steward who maintains definitions and coordinates quality follow-up. Producers explain how source data or transformations work; consumers describe where data does not match an expected use. An organization may combine or assign these roles differently.

To raise a concern, name the dataset and field, describe the observed value or pattern and relevant context, and explain why the mismatch matters. Use an approved internal route and avoid including unnecessary sensitive information.

Who does what

Responsibilities are distinct even when one person covers more than one role.

Dataset owner

Accountable for the dataset’s intended meaning and use, and for decisions about its business definition.

  • Clarify what the dataset is intended to represent.
  • Assess proposed changes to meaning or use.

Data steward

Coordinates definitions and quality follow-up, connecting consumers with the owner or producer when needed.

  • Keep field definitions understandable and consistent.
  • Help route a specific concern to the relevant role.

Data producer

Explains the source and transformation behavior that produces the dataset.

  • Describe relevant source or transformation context.
  • Identify when a proposed change affects data production.

Data consumer

Uses the dataset and can describe a mismatch between the data and its documented meaning or intended use.

  • Compare the observed data with the available definition.
  • Provide a specific, appropriately limited example.

Fictional example: Service Orders

Illustrative sample only; these are not organization-wide definitions or assignments.

Reference item Northstar Works example
Dataset purpose Track service orders from recording through completion.
Owner role Service Operations
Steward role Data Steward
Opened date The date the service order is recorded.
Completed date The date work is marked complete.

Check and raise a possible quality issue

Share enough detail to make the question investigable without asserting the data is definitely wrong.

  1. 1

    Compare with the definition

    Check the field’s stated meaning and whether the value appears consistent with it.

  2. 2

    Add relevant context

    Note the affected period, source, or other context that could explain the pattern.

  3. 3

    Describe one specific example

    Name the dataset and field, explain what you observed, and exclude unnecessary sensitive details.

  4. 4

    Use the established route

    Send the concern through your organization’s verified data-quality route and state why the mismatch matters.

  5. 5

    Coordinate review

    The steward can bring questions about meaning to the owner and questions about source behavior to the relevant producer. Review does not by itself imply a guaranteed response time or automatic correction.

Make the question useful

A clear report distinguishes an observed pattern from a conclusion about its cause.

Specific and investigable

A consumer notices a service order marked complete before its recorded opening date.

  • Name Service Orders and the opened date and completed date fields.
  • Describe the affected record in an approved internal channel, without unnecessary sensitive details.
  • Ask whether the dates follow their documented definitions.

Too vague to route

“The dashboard is wrong” does not identify the dataset, field, observed pattern, or question.

  • Add the dataset and field names.
  • Describe what you observed and the relevant context.
  • State what you need clarified rather than assuming a cause.

How definition and source changes move forward

Raise a proposed definition or source change with the data steward through your organization’s established route. The steward brings questions about intended meaning to the dataset owner and questions about source or transformation behavior to the relevant producer. Those roles assess the proposal and communicate the agreed outcome through the established route.

For a quality concern or change request, use your organization’s verified data-quality channel. If you are unsure which route applies, ask your manager or data and technology team to identify the appropriate internal route.