Data governance fails when it behaves like compliance instead of a product

September 16, 2026

Data governance can be technically correct and still fail.

The policies exist. The standards are documented. The council meets. Data stewards have been nominated. The catalogue has been bought. Training has been delivered. Yet the business still works around the process, ownership remains vague and the same quality issues continue to reappear.

Recent UK data-leader discussions convened by The Leadership Board suggest a reason: governance struggles when it is designed primarily as enforcement. The more interesting operating models are moving toward something closer to product management, where adoption, usability, ownership, feedback and measurable business value matter as much as policy compliance.

For data vendors, that is an important shift. Buyers do not only need governance capability. They need governance that people will actually use.

The old governance model assumes compliance creates behaviour

Traditional governance programmes often begin with rules. Define the policy, assign the role, document the process, create the control and expect the organisation to comply.

That approach is necessary in regulated environments, but it is rarely sufficient. Business teams are already under pressure to deliver revenue, service customers and meet operational deadlines. If governance appears as an additional layer of administration, the organisation creates a predictable response: people comply when forced, work around the process when possible and treat the data team as the owner of the problem.

Senior leaders repeatedly described the cultural difficulty of engaging time-poor stakeholders and data stewards. The challenge was not a lack of policy. It was making the governance work relevant enough that people saw a reason to participate.

Governance is not embedded because a policy exists. It is embedded when the governed behaviour becomes the normal way the organisation works.

A product mindset changes the question

A compliance mindset asks: what rules must people follow?

A product mindset asks: what problem are we solving, who is the user, what behaviour do we need, what makes adoption easier and how will we know the operating model is creating value?

That distinction surfaced explicitly in the roundtables. Data leaders discussed treating governance as a product, with roadmaps, user needs, technical enablement and business outcomes. Some described governance roles evolving toward product-oriented responsibilities that bridge business and technology rather than remaining purely custodial or policy-based.

This does not mean abandoning control. It means designing control around use.

Central policy and local ownership are not opposites

One of the strongest operating-model signals was the movement toward federated governance: central standards with decentralised execution and accountability.

The logic is practical. Enterprise definitions, regulatory controls and minimum standards need consistency. But a central data office cannot realistically own the meaning and quality of every dataset across Finance, HR, Marketing, Operations and specialist business functions.

The business understands the context. The governance function provides the framework. Technology makes the framework easier to follow and easier to evidence.

This connects directly to the ownership gap that blocks enterprise data budgets. When ownership is ambiguous, problems escalate sideways. When ownership is named and supported by usable workflows, the organisation has a route to decision and remediation.

The data steward cannot be a ceremonial role

Many organisations have learned that simply naming data stewards does not create stewardship.

People need to understand what the role means, what authority they have, how much time it requires, where issues are raised and what support exists when a problem crosses functions. Without that operating detail, stewardship becomes an extra title attached to an already busy person.

The discussions included practical mechanisms such as monthly steward forums, data health checks, conceptual modelling workshops, issue-management workflows and cross-functional governance groups. The important point is not that every enterprise should copy the same structure. It is that governance needs an interaction model.

A product team would never launch software and assume adoption because users were told to use it. Data governance should not make that assumption either.

Good governance makes the secure and correct path easier

The most valuable governance technology increasingly disappears into the workflow.

Classification should appear where data is created. Ownership should be visible where issues are raised. Approved definitions should be available where metrics are used. Access requests should route automatically. Sensitive-data controls should follow policy without requiring users to understand the entire governance framework.

This is where vendors can materially change adoption. A catalogue that requires people to leave their workflow, learn a new taxonomy and complete complex forms may satisfy a technical requirement while failing the user. A governance platform that integrates with the systems people already use can create the same control with far less resistance.

The same principle applies to data literacy. As discussed in why data literacy is a hidden gatekeeper to enterprise platform spend, adoption is not solved by access alone. People need enough context to use the platform and understand the consequences of their decisions.

Governance becomes stronger when feedback changes the product

Product teams study behaviour. Governance teams should do the same.

One discussion described using query history and user feedback to identify where certified metrics were not meeting real business needs. If users repeatedly ask for a different definition, that may indicate poor behaviour. It may also indicate that the approved definition is not fit for the use case.

That is a valuable distinction. Governance should not automatically treat every workaround as non-compliance. Sometimes the workaround is evidence that the operating model is failing the user.

A product-minded governance function captures those signals, distinguishes legitimate variation from uncontrolled duplication and updates the roadmap accordingly.

AI makes governance adoption even more important

AI increases the urgency because it dramatically lowers the barrier to data use.

Employees can now interrogate datasets, create local agents, combine information and generate analysis without waiting for a traditional reporting cycle. That is valuable, but it also means weak governance can spread faster.

The answer is not to force every interaction through a central data team. Buyers are looking for ways to keep standards central while allowing local experimentation inside controlled boundaries. That is why semantic layers, sandboxes, role-based access, approved data products and clear ownership are becoming strategically important.

Our analysis of what enterprise teams demand from governance, lineage and metadata reflects the same direction. Governance increasingly has to enable faster data use while making the boundaries more visible.

What product-minded governance looks like

Compliance-first governance	Product-minded governance	What vendors should enable
Policy is the output	Adopted behaviour is the output	Workflow integration and measurable usage
Central team owns governance	Central standards, distributed accountability	Role clarity, ownership and federated controls
Training is generic	Education is role and context specific	Embedded guidance and targeted learning
Issues are escalated manually	Issues have visible product workflows	Routing, prioritisation and audit trails
Definitions are fixed centrally	Definitions are governed but evolve with need	Semantic management, feedback and change control
Success means policy coverage	Success means trusted reuse and business value	Adoption, quality and outcome metrics

The commercial mistake is selling governance as restriction

Governance vendors often lead with risk reduction because risk is easy to understand and compliance creates urgency. But a proposition built only around restriction can reinforce the buyer’s cultural problem.

The stronger story is controlled enablement. Better governance should make it easier for the business to find trusted data, understand what it means, request access, resolve issues and reuse approved assets. It should reduce the uncertainty that causes central teams to say no.

That is especially important when the buyer is trying to scale AI. The organisation does not want a governance programme that slows every use case. It wants an operating model that tells people what they can do, what needs approval and how to move safely from experimentation to production.

What data governance vendors should change

Sell adoption as a governance outcome

Do not stop at policy coverage, catalogue completion or control counts. Show how the platform increases trusted reuse and reduces workarounds.

Design for the steward who has another job

Assume business owners are time poor. The workflow must make participation easier, clearer and faster than ignoring the process.

Prove how central control and local autonomy coexist

Enterprise buyers need consistency without creating a central bottleneck. Demonstrate federated permissions, local ownership and enterprise-wide standards in the same operating model.

Treat feedback as governance data

Show how the platform captures usage, failed searches, unresolved issues and recurring requests so the governance programme can improve rather than simply enforce.

Connect governance to business value

Help buyers show how trusted data shortens decisions, reduces rework, enables AI and protects high-value processes. Governance becomes easier to defend when it is seen as an operating capability rather than overhead.

Governance has to earn adoption

Enterprise data governance will always contain rules, controls and regulatory obligations. The mistake is assuming those things automatically create a healthy data culture.

The more durable model treats governance as something that has users, workflows, feedback, outcomes and a roadmap. It gives central teams control where consistency matters and gives the business ownership where context matters.

For vendors, that changes the pitch from “we help you enforce governance” to “we help governance become the way good data work gets done”.

The Leadership Board gives technology vendors access to senior enterprise buyers while these operating models are still being designed. If your proposition makes governance easier to adopt, easier to evidence and more useful to the business, the strongest sales conversation starts before the framework is locked in.