Proxideo Partners
About Services Leadership ETRIS Blog Contact Engage Us
← Back to Insights

Data Governance Isn't an IT Problem. It's a Board Problem.

For two decades, organizations handed data governance to IT and called it a day. That logic doesn't work anymore, and the costs of ignoring it are compounding.

What is data governance?

Data governance is the system of decision rights, accountability structures, policies, and standards that determine how an organization manages its data assets, including who owns data, who can access it, how quality standards are set and enforced, and who is accountable when those standards are not met. It is not primarily a technology function; it is an organizational and leadership function that requires executive authority and, increasingly, board-level oversight. Poor data governance manifests as data quality failures, regulatory exposure, fragmented ownership, and AI programs that cannot produce reliable outputs because the underlying data cannot be trusted.

Data Governance and IT: Twenty Years of Misplaced Accountability

For two decades, organizations handed data governance to IT and called it a day. The assumption was simple: data lives in systems, IT manages systems, therefore IT should govern data. That logic made sense in a world of mainframes and monolithic databases. It doesn't work in a world of cloud platforms, data lakes, AI, and regulatory scrutiny. And yet, most organizations haven't moved past it. They've just added more dashboards and tried to "portal" their way out. The same structural inertia shows up in enterprise transformation programs broadly: accountability gaps that leadership assigns to the wrong level tend to stay there until a crisis forces the issue.

The result is predictable, well-documented. Data quality problems. Regulatory gaps. Inconsistent definitions across business units. Fragmented ownership. And governance frameworks that never quite stick. Frameworks don't fail due to bad architecture. They fail when the people accountable for outcomes don't have the authority, the incentive, or the organizational mandate to enforce anything.

Data stewardship stays lip service when it lacks executive air cover. Business units can define data domains but without board-level accountability, there's no forcing function for alignment. The result: governance becomes a compliance checkbox, not a strategic function.

Here's the thing: IT teams aren't failing at data governance because they're incompetent. They're failing because they were never supposed to own it in the first place. Governance is about decision rights, accountability structures, and organizational behavior. It's a management problem. It's a leadership problem. And when you push it down into infrastructure teams, you get infrastructure-shaped answers to questions that are fundamentally about strategy.

If data is the "new oil" on which our world will run, then data should be treated with the respect corresponding to its strategic importance.

The organizations that get governance right don't start with tools. They start with authority. Who owns this data? Who decides what "good" looks like? Who is accountable when something goes wrong? Those aren't IT questions. They're board-level questions. And until a board is willing to treat data governance with the same seriousness it treats financial governance, nothing meaningful will change. The dashboards will keep getting prettier, but the underlying dysfunction will remain.

This isn't a technology problem anymore. It hasn't been for a while. The dynamic changed when data became the operating system for decision-making.

The Real Cost of Poor Data Governance: $12.9M Per Year and Counting

Most organizations have no idea what poor data governance is actually costing them, because the costs bleed across multiple budget lines and never get added up. Gartner pegs the average cost of poor data quality at $12.9 million per year, and that probably understates it for large enterprises with complex environments. (Source: Gartner, 2023)

The direct costs are real enough: regulatory fines when data failures trigger compliance violations, AI projects that collapse because models trained on garbage data produce garbage outputs, financial reporting corrections, operational errors from decisions made on data that doesn't reflect reality. But the indirect costs are where things get interesting. Organizations that can't trust their data pay a decision velocity penalty. They slow down because they can't act confidently on what they're seeing. Competitors with cleaner data deploy AI and analytics capabilities that data-challenged organizations simply can't match. There's also the hidden operational tax: all the hours burned on data reconciliation, exception handling, and manual correction that wouldn't be necessary if quality had been maintained upstream.

That last one is death by a thousand cuts. Most leadership teams never see the bill.

Board-Level Data Governance Failures: Case Studies Every Director Should Know

You don't have to speculate about what board-level data governance failures look like. Several major financial institutions have faced regulatory enforcement actions in recent years, in part because their data quality and data lineage capabilities couldn't satisfy reporting requirements. The Federal Reserve and OCC have cited data governance deficiencies explicitly in consent orders, in some cases imposing restrictions on growth and dividends until remediation was complete. The financial hit and the reputational fallout were both board-level problems.

AI failure is the newer category, but it's becoming familiar. A program that burns through $15 million in development and never reaches production because the training data was too inconsistent to produce reliable outputs. That's not an IT failure. Before committing to any significant AI investment, data governance readiness should be a scored precondition. The board approved that AI budget. The board should have set data quality as a precondition. It didn't.

And then there's competitive disadvantage, which is quieter but cumulative. You can't deploy personalization at scale if your customer data is a mess. You can't close books fast if your financial data won't reconcile across systems. You can't build accurate AI demand forecasting on fragmented product data. These disadvantages don't show up in a single bad quarter. They compound over years.

What Effective Board-Level Data Governance Looks Like in Practice

A board that takes data governance seriously operates differently from one that delegates it down. Five things distinguish the serious approach.

Start with a data asset inventory and materiality classification. The board should know which data assets are strategically critical, which carry the highest regulatory risk, and which are prerequisites for the organization's AI strategy. This is not a technical document; it's a strategic asset register, analogous to what you'd maintain for physical assets.

Second, build an executive-level ownership structure. Every critical data domain (customer data, financial data, product data, operational data) needs a named executive owner who is accountable for quality in that domain. Not an IT data steward. A line executive with real organizational authority over how that data is created, maintained, and used.

Third, put a data quality dashboard in front of the board. Apply the same scrutiny to operational and customer data that you apply to financial reporting. A quarterly data quality report creates the accountability and visibility that actually drives improvement.

Fourth, fund it like a strategic priority. If data governance matters, the investment should reflect that. Squeezing it into an IT operational budget signals that it doesn't really matter.

Fifth, link AI readiness directly to data governance outcomes. When an AI program fails because of data quality problems, that failure should land at the board level as a data governance failure, not get buried in an IT postmortem.

"Governance frameworks don't fail due to bad architecture. They fail when the people accountable for outcomes don't have the authority, the incentive, or the mandate to enforce anything."

Why Moving Data Governance to the Board Level Actually Changes Behavior

The boards that have made this transition describe a consistent experience: once data quality becomes a board-level accountability, behavior changes fast. Executive data owners can't ignore quality problems the way IT stewards could. They have too much skin in the game. Budget requests for data infrastructure remediation get a different hearing when the board has explicitly tied data quality to strategic results. And AI programs built on clean, well-governed data succeed at rates that programs on ungoverned data simply don't. This is especially true as agentic AI systems take on more autonomous decision-making: the quality and lineage of data becomes a governance and liability issue, not just an operational one.

None of this is technically complicated. It's organizationally significant. It requires the board to accept that data governance is within their oversight responsibility, not something to be delegated away and forgotten. The organizations that have crossed that line are better positioned for AI, better positioned for regulatory scrutiny, and better positioned for the data-intensive competitive environment that's coming. The ones that haven't crossed it are one crisis away from wishing they had.

Key Takeaways

Frequently Asked Questions

Why should the board of directors be involved in data governance?

The consequences of data governance failures are board-level in nature: regulatory enforcement actions with material financial penalties, restrictions on growth and dividends, failed AI investments, reputational damage, and sustained competitive disadvantage. When governance accountability is pushed down to IT, there is no organizational forcing function. Frameworks become compliance checkboxes rather than strategic functions. The board accepting explicit oversight responsibility for data governance is what creates the accountability dynamic that actually changes behavior across the organization.

How does poor data governance affect enterprise AI programs?

AI systems produce outputs that are only as reliable as the data they are trained and operated on. Poor governance, inconsistent definitions across business units, missing critical fields, fragmented ownership, and uncontrolled data quality, are some of the primary causes of enterprise AI program failure. An organization that approves a multi-million dollar AI investment without first establishing data governance as a precondition is setting up a foreseeable failure. When an AI program collapses because of data quality problems, that failure should be attributed at the board level, not buried in an IT postmortem.

What does board-level data governance look like in practice?

Five elements distinguish a serious approach. First, a strategic data asset inventory that classifies data by criticality and regulatory risk. Second, named executive owners for each critical data domain: customer, financial, product, and operational, with real organizational authority. Third, a data quality dashboard presented to the board on a regular cadence. Fourth, dedicated funding that signals governance is a strategic priority. Fifth, an explicit policy that AI program approval is conditioned on data governance readiness in the relevant domains.

What are the regulatory consequences of inadequate data governance?

Significant and escalating. The EU AI Act, GDPR, and sector-specific regulations, HIPAA in healthcare, FINRA and banking regulators in financial services, all create direct regulatory exposure for organizations that cannot demonstrate adequate data governance. Financial regulators have cited data governance deficiencies explicitly in consent orders, in some cases restricting organizational growth and dividend payments until remediation is complete. As AI regulation matures globally, data quality, lineage, and auditability are increasingly treated as preconditions for regulatory compliance, not optional best practices.

KC
Kevin Carl
Founder & Principal Advisor, Proxideo Partners

Kevin Carl is the Founder and Principal Advisor at Proxideo Partners. He has led or advised more than 20 enterprise transformation programs across travel and hospitality, professional services, and technology-enabled businesses. He was named one of “The Top 25 Artificial Intelligence Consultants and Leaders of 2024” by Consulting Report magazine. Previous roles include Managing Director at turnaround giant Alvarez & Marsal, General Manager and SVP of Engineering at tech unicorn Copado, Group Vice President of Global Hospitality Consulting at Oracle, EVP & Global CIO at Radisson Hotel Group, and Global Managing Director of Digital Travel at Accenture.

Related Reading
AI Strategy
The Agentic AI Shift: What Happens When AI Systems Start Managing Other AI Systems
Transformation
Why Most Enterprise Transformations Fail, and What the Programs That Succeed Do Differently
AI Strategy
Is Your Organization Actually Ready for AI? 6 Questions Every Executive Should Answer Before Writing a Check
AI Strategy
From Pilot to Platform: The Executive's Guide to Scaling AI Across a Complex Organization
Stay Informed

Executive insights,
delivered directly.

Practical perspectives on transformation, AI strategy, and leadership, written for senior executives navigating high-stakes decisions.

We respect your privacy. Unsubscribe at any time.