Poor data governance not just embarrassing, it’s expensive

While leaders express confidence in their own data to fuel their AI initiatives, they have also experienced material financial losses as a result of poor data governance, but this has not blunted efforts to implement AI at their organization, nor has it impacted plans to invest heavily in the technology. 

Processing Content

A recent poll from corporate performance management solutions provider OneStream found 79% of executives believe their data governance can support large-scale AI adoption. However, further data shows they’re not actually as confident as they say, as 61% said they second-guess their data at least once a month, and 11% question it daily. 

They are probably right to question their data’s quality, as 72% say bad data cost their organization $500,000 or more, with more than one-third (37%) reporting damages over $1 million. Downstream impacts include delayed reporting and closing (cited by 44%), lost revenue opportunities (41%), a lack of trust in automated insights (38%) and compliance issues (35%). 

Pam McIntyre, OneStream’s chief accounting officer, said in a later email that by “bad data” they mean inaccurate data but, more broadly, data that is disconnected, opaque, inconsistent, outdated or poor quality. Any one of those characteristics can undermine a decision, she said; it’s not just about things being “right” or “wrong.” This is especially the case if, as many organizations are doing, they are implementing and scaling AI based on said “bad data.” 

“We found a strong relationship here, and that alone points to significant governance gaps that organizations need to address. If companies are adopting AI tools at scale without fully trusting or governing the data those tools rely on, it’s not surprising that poor data-driven decisions follow. That said, AI isn’t so much the root cause as it is an amplifier of underlying data challenges that already existed. The correlation should serve as a clear signal to act, not a verdict on AI itself,” said McIntyre. 

While the vast majority of executives, 85%, say they have a formal data governance program in place or underway, McIntyre said it doesn’t mean much if they lack confidence in the data itself. The challenge is not so much having a governance program but ensuring the data is genuinely auditable and transparent, which she said requires human logic. This, in turn, is needed to establish human trust.  

“Systems don’t build trust on their own. It’s up to humans to create the data hierarchy and define the business logic. You need a person behind the numbers to establish that framework and understand the ‘why,’ which ensures the data is clean and transparent enough for AI to be successfully built on top of it,” she said. 

In the same regard, while executives regularly doubt or double-check the data, this means little if this does not, in turn, turn that skepticism into meaningful investigation. While recognizing a gap is a starting point, she said, it needs to be followed by deeper research, allowing organizations to truly understand the root of their data before reacting.

One of the challenges of establishing a decent data governance program is that organizations do not always agree on who should be responsible for it: finance or IT. The OneStream survey found that while 89% of executives overall say finance and IT are aligned, 85% of CIOs believe they lead data governance, while 78% of CFOs claim the lead. They also have a different perspective on data governance challenges: Finance prioritizes accuracy, context and accountability, and IT focuses on enablement, scalability and execution. Because of this disconnect, nearly one in three CFOs (32%) cite “lack of data ownership” as a key barrier to success. 

This calls to mind a recent study from Grant Thornton that found similar mismatches between different parts of the organization (see previous story). GT found that, within the C-suite, 39% of CIOs and CTOs said their workforce is fully ready to adopt AI compared to only 7% of COOs. Further, the study found that 54% of COOs are concerned about regulatory and compliance uncertainty related to agentic AI, versus 20% of CTOs.  

A similar confidence mismatch was found in a recent survey from data extraction solutions provider Parseur, which noted that, on the one hand, 88% of U.S. business leaders said they are very or somewhat confident in the accuracy of the data feeding their analytics and AI systems but, on the other, 69% of them reported discovering errors in document-derived data at least sometimes, if not often or very often. Parseur said this implies finding mistakes in the data is a regular occurrence. The study also pointed out that the quality of the AI models that everyone seems to be using today depends on the quality of the data fed into them, which implies that overconfidence in data quality is a risk factor for AI implementation. 

This confidence mismatch was also found in Deloitte’s State of AI in Enterprise report, which found that a comfortable 74% majority of companies said they plan to deploy agentic AI sometime within two years, 23% are already using agentic AI at least moderately, and 85% of companies expect to customize agents to fit the unique needs of their business. Trust and investment in AI generally is also on the rise, with 84% of organizations increasing their AI investments, and 78% of leaders reporting greater confidence in the technology. At the same time, only 21% of companies say they currently have a mature model for governance of autonomous agents.

However, governance lagging behind ambitions in AI is not a new thing. Last year, for example, governance, risk and compliance solutions provider AuditBoard (now Optro) found over 80% of respondents said their organizations are either very or extremely concerned about AI risks but, at the same time, only 25% said they have fully implemented an AI governance program. Meanwhile, although 92% of respondents said they are confident in their visibility into third-party AI use, just 67% of organizations report conducting formal, AI-specific risk assessments for third-party models or vendors. That leaves roughly one in three firms relying on external AI systems without a clear understanding of the risks they may pose.