Why Data Governance Fails When Only IT Is In The Room

Jacqueline DeStefano-Tangorra is the president and CTO of DataOps.

getty

In the last few years, across more than 200 engagements spanning defense, manufacturing, retail, aerospace, construction and supply chain, I've had one conversation more than any other: A leadership team wants to deploy AI, they have budget, they have executive buy-in, but their data is a mess.

Typically, it's not because they lack tools but because nobody outside of IT was ever asked to participate in the governance of core data. Gartner Inc. predicted in February 2024 that 80% of data and analytics governance initiatives would fail by 2027 due to crisis-driven urgency, and that number doesn't surprise me because I've watched it play out in person.

However, it's not because people are resisting governance. It's because the right people were never in the room for the conversations that shaped the governance program in the first place.

IT is necessary, but it's truly not enough anymore.

IT is responsible for managing the infrastructure, access controls and pipelines, but data governance can't live inside one department if you truly expect enterprise-wide adoption. MIT Sloan Management Review cited a 2020 NewVantage Partners survey of Fortune 1000 executives, which found that 90% noted people and process issues as the principal barrier to becoming data-driven, while only 9.1% pointed to technology.

I've seen this firsthand more times than I can count. The platform is solid, and the architecture is sound, but the business still doesn't trust the data. Trust doesn't get inherited from the tech stack. It gets built through cross-functional involvement, shared definitions and accountability that extends beyond the technical team.

A 2021 Drexel University survey found that nearly two-thirds of responding organizations reported cultural awareness and adoption as the top obstacles to data governance. When governance is built without the people who touch the data daily, adoption becomes an uphill battle that most organizations lose.

This work requires someone who understands both sides.

The engagements where I've seen global data governance take hold share something in common. The person leading it can sit with a data engineer and understand pipeline logic and then walk into a boardroom and explain why that pipeline and governance process matters to revenue. They understand that a procurement team needs different guardrails than an engineering team. They know that if governance doesn't map to how people already work, people will find a way around it.

A 2011 Gartner survey found that companies were allocating on average only 5% of implementation budgets to change management, while Gartner recommended 15%. That tells you everything about where the disconnect lives.

The stakes are getting higher.

Gartner research from 2020 found that the average organization was losing $12.9 million annually due to poor data quality. That number gets harder to absorb as enterprises move toward agentic AI, where systems make decisions autonomously based on the data they are given. Ungoverned data in that context creates much more than inefficiency. It creates organizational risk at scale.

Gartner also predicted in February 2025 that organizations would abandon 60% of AI projects through 2026 due to insufficient data quality. The companies that avoid that outcome will be the ones that treat governance as something the whole organization owns, long before the first model was ever deployed.

Where do leaders need to start?

Bring business unit leaders into the data governance conversation from day one. Let them define what data matters and why. In my experience, governance gains traction when business leaders see themselves as co-owners of the data, not just consumers of reports that IT produces from it.

Start narrow. Focus on 10 to 15 critical data fields and core KPIs with clear data ownership and weekly reviews, then expand once value is demonstrated. You don't need every country or business unit on board from day one. You need one validated use case that works, in one region, with one team that can speak to the results. The rest will scale from there.

Pair that with a change management investment that matches the scope of what you're asking people to do. If you're asking an entire organization to change how it interacts with data, a one-time training session and a policy document won't get you there. Governance adoption requires ongoing communication, feedback loops and visible executive sponsorship.

Measure adoption, not just compliance. According to Infoverity, lack of adoption is the clearest sign of governance failure. If people are working around your framework, listen to that signal. It's telling you that governance was designed for the system, not for the people using it.

Data governance is how an organization builds the foundation for everything it wants AI to do. That foundation has to belong to the business, not just the team managing the servers.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?