Dun & Bradstreet survey finds few enterprises have AI-ready data
Only 8% of 10,000 surveyed businesses said their data was ready for AI, and reported returns varied sharply by readiness level.
By Dominic Okoye · Staff Writer
· 3 min read
Enterprise AI readiness and ROI remain far apart for many companies, according to a Dun & Bradstreet survey of 10,000 businesses in 32 countries reported by CIO Dive. Just 8% described their data as ready to support AI, while companies reporting fully AI-ready data were far more likely to report strong returns than those with only partially ready data.
The survey found that 33% of respondents characterized their data as mostly ready for AI and 44% as partially ready. Among companies with fully AI-ready data, 81% said they were seeing strong AI returns, compared with 12% of companies whose data was partially ready. Those figures describe a self-reported association, not proof that data readiness by itself produces financial returns.
Gary Kotovets, Dun & Bradstreet's chief data and analytics officer, said the quality of information available to models will matter more in the next stage of adoption than access to models. CIO Dive also reported that businesses in the Americas were more likely to raise AI spending while reporting the second-lowest rates of strong ROI and mostly or fully AI-ready data.
Why does AI readiness affect enterprise ROI?
Data readiness is a practical operating question rather than a single technical benchmark. Frank Iannella, Heineken's U.S. CIO and senior vice president for digital technology, described decision-ready data as current, good enough to support action and connected to the business levers for the relevant use case, according to CIO Dive.
That distinction matters because an employee's reported time savings does not establish an enterprise-level return. KPMG Canada said its November 2025 survey found that 93% of Canadian business leaders were using or piloting AI, but only 2% reported measurable ROI. It also said 31% of organizations had embedded generative AI into core operations and workflows.
Employee reports look different. KPMG said 79% of Canadian employees who use generative AI at work reported improved productivity. That can be a useful input, but it is not the same measure as a company-wide financial return, which depends on whether the organization can translate saved capacity or improved output into business results.
Data is one constraint, not the whole answer
KPMG identified workforce literacy, workflow integration, organizational guidance and governance as additional gaps between adoption and sustained value. Its 2025 index found 83% of employee AI users believed they needed to improve their skills, while 48% said their employer offered sufficient role-specific training. Only 29% said their employer had a comprehensive policy on acceptable AI use cases, and 42% were unsure whether one existed.
Oliver Wyman Forum argues that AI returns can be delayed, dispersed across an organization or dependent on changes outside the technology itself. The consultancy also says AI can shift bottlenecks instead of eliminating them, requiring workflow redesign and executive accountability to capture gains. It reported that workplace AI users told it they spend 22 minutes checking, correcting or redoing AI output for every hour of AI-assisted work; the excerpt does not specify the underlying methodology.
For operators, the surveys point to separate questions that should not be collapsed into one ROI claim: whether data can support a use case, whether AI is embedded in the workflow, whether staff can use it reliably and whether the business has a way to measure and capture the resulting value. Targeted, foundational, workforce and transformational AI investments can have different paths to return.
This story draws on original reporting from CIO Dive.
