Data governance doesn't decorate. It decides.
Clear roles, differentiated custody, traced access, governed initiatives. A decision system over data that operates every day — not a document updated once a year.
- 3 models: centralized → distributed → federated
- 7 roles with explicit decisions
- Differentiated custody at 3 levels
- Governed cross-domain access workflow
- Single corporate initiatives backlog
Data-Driven isn't having data. It's deciding with it.
A data-driven organization is one where every meaningful decision — operational, commercial, strategic — relies on governed, accessible and trustworthy data. Data governance is the precondition, not the result.
Data is defensible and traceable. Every KPI has origin, formula, purpose and owner. Business trusts data without re-validating it each time.
Frictionless, without going through IT. Governed publication lets each profile reach the data they need within the authorized perimeter.
Business decides meaning, expected quality and correct use. Functional accountability lives in the domains, not in the platform.
Deciding with data is the norm, not the exception. The organization learns to defend decisions by data — not intuition or pressure.
Three governance models. An evolutionary sequence. No skipping stages.
Datum rolls out governance in three consecutive phases. Each builds on the previous one. No organization jumps straight to federated without consolidating distributed first. None reaches distributed without stabilizing centralized first.
All governance operations are managed and moderated by the corporate Data Committee. Framework, policies, domains, quality and decisions concentrate to stabilize the model and build initial trust.
Priority shifts to functional users. Active Owners and Stewards are consolidated by domain, with a specific committee that decides within its scope. The corporate Committee coordinates but no longer operates the detail.
Each domain has its own authority to define, publish and evolve its data within the common framework. The corporate Committee only supervises and resolves cross-cutting or regulatory exceptions.
Data governance isn't documented. It's operated.
Governance that doesn't decide access, doesn't prioritize initiatives and doesn't measure itself is just a PowerPoint with a corporate seal. Datum turns it into an operating system.
Four frictions that separate written governance from governance that operates.
It isn't a methodology problem nor a tooling problem. It's an operating model problem.
The role exists on paper but does not decide policies, does not approve domains and has no mandate to resolve conflicts between areas. Governance stays as recommendation.
They appear on an org chart but do not validate quality, do not approve access and do not assume the functional meaning of data. Functional responsibility returns to IT.
Infrastructure, platform and functional data live mixed together. IT ends up deciding what business should decide.
When a single backlog and objective prioritization criteria are missing, demand enters through the loudest channel. The data roadmap is built reactively — DATUM brings corporate backlog and signed criteria.
Seven roles with explicit responsibilities, decisions and limits.
Separation of responsibilities removes the common pattern of treating as technical what really requires functional business decisions. The collegiate bodies (Data Office and Data Committee) appear later in this page.
Strategic data owner. Defines corporate direction and aligns business, technology, security and exploitation.
Day-to-day operation of governance. Implements policies, coordinates domains, follows up and prepares the Committee.
Technical-functional data design: layers, flows, patterns and model. Ensures governance is executable.
Technical custody of the platform. Configuration, performance, automation and technical security of the how.
IT, Cloud, Systems. Custodies infrastructure: connectivity, secrets, identity, network and underlying compute.
Functional ownership of the domain. Concepts, correct use, expected quality, domain KPIs and validation of initiatives.
Operational management of data within the domain. Coherence, incidents, documentation and correct use day to day.
Three custody levels separated. Mixing them is the most common cause of failure.
Functional custody is exercised by business. Platform custody by architecture. Infrastructure custody by IT. Three responsibilities, three custodians.
Meaning, expected quality, domain ownership and use validation. Exercised by the Data Office over Owners and Stewards.
Data model, live metadata, processing engines, automation and technical security of the platform.
Connectivity, secrets, identity, network, storage and compute. Authorized connections without exposing credentials to DataOps.
Five steps for cross-domain access to be legitimate, justified and traceable.
Being a Data Owner doesn't grant access to other domains. Every cross-domain access relationship goes through a governed, registered and reviewable process.
The consuming Owner requests access stating purpose, term and use. The source Owner validates compatibility with the functional ownership of the data.
The Data Office reviews coherence with policies, least privilege and corporate publication perimeter.
The Data Committee steps in when there is cross-cutting, regulatory or significant semantic impact. Otherwise the Office handles it.
Architecture and SMEs implement the approved access through the appropriate technical mechanism. Business never enables directly.
The platform records and versions the grant. The consuming Owner and the Data Office periodically review that the need is still valid.
Being a data owner doesn't grant access. Governance does.
Every cross-domain access goes through a chain of validation that protects functional ownership, purpose of use and the principle of least privilege. Traceability isn't optional.
Two collegiate bodies that turn governance into real operation.
The Data Office and the Data Committee aren't individual roles — they are corporate bodies with their own composition, attributions and cadence. Together they materialize governance as operating capability.
Data Office
Defines and maintains corporate data policies, standards, procedures and metadata.
Oversees scorecards, quality rules and incidents. Records and manages the Corporate Data Backlog.
Coordinates Data Owners and Data Stewards across domains. Prepares, documents and follows up on the Data Committee.
Data Committee
Approves corporate policies, standards and exceptions. Validates domain assignment and main Data Owners.
Approves corporate KPIs and their calculation criteria. Decides on cross-domain access with regulatory or semantic impact.
Prioritizes and approves data initiatives with cross-cutting impact or that require corporate decision.
A single repository where all data demand is registered, prioritized and traced.
Prioritization applies objective criteria: business impact, risk, regulation, complexity and strategic alignment — not arrival order or one-off pressure.
A stakeholder, domain, technical area, governance or compliance team proposes the initiative to the backlog.
The Data Office registers the initiative in the Corporate Backlog, assigns an identifier and a responsible stakeholder.
Functional, technical, quality, security, architecture and cost analysis by the Data Office.
The Data Committee evaluates when the initiative is cross-cutting or requires corporate decision. Otherwise the Office resolves it.
Approval, rejection or request for additional analysis. Assignment to executing area with implementation roadmap.
Follow-up by the Data Office until closure, with continuous Committee visibility and full traceability.
Four capabilities operational governance gives to data — and to the organization.
And one cross-cutting capability: ethical use of data. Legitimate purpose, proportionality, transparency and human oversight in every decision.
Every decision over data has origin, context and record. The organization can explain why it accessed, why it prioritized and why it published.
The corporate backlog and prioritization criteria accelerate initiatives instead of slowing them down. Speed comes from order, not chaos.
When Owners decide meaning and KPIs are validated by the Committee, business trusts data without re-validating it every time.
The distributed model lets new domains onboard without breaking the corporate framework. Governance grows with the organization.
A specific profile, a real question.
Hay políticas, hay roles, hay comités. Pero los metadatos no se actualizan, la calidad no se mide de forma sistemática y la trazabilidad se reconstruye manualmente cuando alguien la pide.
Gobierno centralizado con Data Owners activos, catálogo vivo y políticas de calidad automatizadas en una entidad bancaria mediana.
Connect with the next layer
To govern is to transform. Here are the three paths to put it into practice.
Data governance is the data's nervous system.
In 4–8 weeks, the operational framework active in your organization: roles assigned, Committee constituted, backlog running, first domain in pilot.