• About us
    About us
    question mark
    Who we are

    Learn more about Mantu values, governance and offices.

    hexagon
    Our brands

    11 brands united by a shared vision.

    leave
    Sustainability

    Our strategy through diversity, environment and innovation.

    bookshelf
    Pressroom

    Breakthroughs, partnerships, and voices behind the transformation.

  • What we do
    What we do
    mantu
    PRACTICES

    Four practices designed to empower organizations, connect talent, and shape sustainable growth.

    cpu
    Technology

    Deep industry knowledge & cutting edge technology to co-create meaningful solutions.

    handshake
    Total Talent Management

    Tech to boost talent and create strong links between companies and the minds they need.

    digital qr
    Creative Intelligence

    Ensure continuity between decision, activation, and adoption. One team, one trajectory, through to lasting impact.

    medal
    Leadership & Advocacy

    Equip executive teams to define their purpose, shape their positioning and drive their strategy.

  • Insights
    Insights
    book open 4
    Blog

    Bold thinking. Fresh perspectives.

    book check
    Client Stories

    Where audacious ideas turn into real stories.

    mantu best managed companies award
    Mantu awarded one of Switzerland’s Best Managed Companies 2025 by Deloitte

    This award highlights the exceptional performance of privately held Swiss companies that demonstrate excellence in strategy, governance, innovation, and long-term results.

    Read more
    WeMeet 2025-2772 1 1
    Mantu signs the DEI Charter

    At the beginning of July 2025, Mantu’s Executive Committee signed the DEI Charter to foster diversity, equity, and inclusion at Mantu.

    Read more
  • Careers
    Careers
    binoculars
    Life at Mantu

    Mantu, as seen by its team members.

    building
    Find a company

    Mantu brings together complementary brands that cover many sectors, all around the world.

Data Ownership in Enterprise Architecture: Roles and Decision Rights

Data Ownership in Enterprise Architecture: Roles and Decision Rights

Data ownership defines who is accountable for enterprise data and who can make decisions about its use, quality and maintenance.

This article focuses on the ownership model behind the chosen architecture.

What data ownership covers ?


In an enterprise setting, ownership means deciding what data means, what quality is required, who can approve changes and who must respond when something goes wrong.

That makes ownership a system of decision rights. A data owner may approve a business definition. A steward may monitor a quality rule and coordinate remediation. A custodian may implement access controls or repair a failed pipeline.

Without this separation, technical teams are asked to make business decisions, while business teams assume that a platform team owns the consequences of their data choices. This connection between business accountability and technical stewardship is central to the wider Data Strategy & Operating Models question.

Data owner, data steward and data custodian


The titles vary between organisations. The responsibilities should still be documented.

Role

Accountable for

Typical responsibilities

Data owner

Business meaning, acceptable risk and priorities

Approves definitions, quality expectations, access principles and material changes

Data steward

Day-to-day data management

Maintains metadata, monitors quality, coordinates issue resolution and supports users

Data custodian

Technical operation and protection

Runs platforms, manages permissions, maintains pipelines, logging, backups and safeguards

The data owner needs authority within the business domain. They do not need to configure a warehouse or fix a transformation job. Their role is to decide what the data should mean and what level of quality is required for its intended use.

The data steward turns those decisions into operating practices. They may maintain a glossary, investigate incomplete records or coordinate a change between data producers and consumers.

The custodian makes the technical controls work. They can identify a failed check or an access issue, but they should not decide whether a business definition is correct.

Citation Quote Primary

Data ownership refers to the accountability and responsibility for the data.

DAIN Studios, The Importance of Data Ownership in Data Governance

The quote captures the basic test for an ownership model: someone must have the authority to decide, and someone must be responsible for the result. A list of names is not enough if those people cannot approve changes, set quality expectations or require remediation.

In a small organisation, one person may hold two or even three roles. The distinction still matters because technical administration does not replace business accountability.

How ownership changes by architecture model ?


For more on the main architecture models, read Mantu’s guide to types of data architectures.

Here, the relevant question is narrower: where does accountability sit once an organisation adopts one of these models?

Architecture model

Where ownership sits

Decision rights to clarify

Main ownership risk

Centralised

Central data function, with business sponsors for key domains

Who approves definitions, access and quality rules

The central team becomes a bottleneck or lacks business context

Federated

Business domains, under shared enterprise standards

Which decisions belong to domains and which remain central

Domains publish inconsistent definitions or quality expectations

Data mesh

Domain teams responsible for data products

Who owns product quality, documentation, contracts and change notices

Teams receive product ownership without the capacity to maintain it

A centralised model concentrates authority. It can suit organisations with limited domain capability or strong consistency requirements, but business sponsors should still own the meaning and use of critical data.

A federated model moves accountability closer to the source. Domains take responsibility for the data they create, while a central function defines shared standards, interfaces and escalation routes.

A data mesh extends this logic to data products. Each product needs a named owner, documented consumers, quality expectations and a technical team responsible for its operation. A platform cannot create those responsibilities by itself.

The important design task is to record the boundary between local and enterprise decisions. A domain may define a local operational measure. An enterprise forum may still need to approve the definition of a shared customer identifier.

Decision rights make accountability operational


Ownership becomes useful when it answers practical questions before an issue occurs:

  • Who approves a new definition?

  • Who decides whether a quality issue blocks a report or process?

  • Who must be notified when a source field changes?

  • Who approves access for a new use case?

  • Who decides whether a data product can change its schema?

  • Who resolves a conflict between two domains?

These decisions should be recorded in an ownership register or responsibility matrix. The document should name the owner, steward and custodian, but it should also state the limits of their authority.

For example, a customer support data product may require fresh contact details during working hours. The custodian implements the check, and the steward monitors it. The owner decides what happens when freshness falls below the agreed threshold, including whether the issue requires remediation, escalation or acceptance of risk.

Data structures and platforms need decision processes that people can actually operate. This is part of the wider data strategy consulting question.

How to implement a data ownership model ?


A practical implementation can follow five steps.

  1. Map critical domains and data products. Start with the data used in important processes, reports, models or customer interactions.

  2. Name owners with decision authority. Avoid assigning ownership to a mailbox or a team with no power to approve changes.

  3. Separate the three responsibilities. Document the business owner, operational steward and technical custodian, even when one person fills several roles.

  4. Record decision rights. Specify who approves definitions, access principles, quality thresholds, schema changes and exceptions.

  5. Measure behaviour. Track unresolved quality issues, time to assign incidents, assets with named owners and changes made without consumer notification.

The ownership register should be tested against real scenarios, not reviewed only as a document. If a CRM system changes the customer_status field, the data owner should approve the new definition, the steward should assess the impact on users and reports, and the custodian should update the technical mapping. If a finance report fails its completeness threshold, the owner should decide whether the report can still be used, the steward should coordinate remediation and the custodian should investigate the pipeline. If these responsibilities cannot be assigned quickly, the register is incomplete.

Review ownership when the architecture, business structure or critical data flows change. A model that worked in a central warehouse may become unclear after domains begin publishing data products across several platforms.

Make ownership part of the architecture decision


Data ownership is the link between business accountability and technical stewardship. It determines who defines the data, who maintains its quality and who can make changes without creating uncertainty for other teams.

The architecture model changes the distribution of responsibility, but every important data asset still needs a named owner, a working stewardship process and a custodian who can enforce the agreed controls.

Organisations reviewing this relationship can discuss their priorities through data strategy consulting, with a focus on ownership, decision rights and the operating model behind the target architecture.