Businesses are increasing their investment in artificial intelligence, but many still lack the data foundations needed to turn pilots into measurable business results. In Dun & Bradstreet’s latest AI Momentum Survey of 10,000 businesses across 32 countries, 65% of respondents plan to increase AI spending over the next 12 months, while just 28% report broad or strong returns from their AI initiatives. Only 8% say their enterprise data is fully ready to support AI at scale, highlighting a gap between investment ambitions and organizational preparedness.
Data Readiness Emerges as a Key Barrier to Enterprise AI Returns
The findings suggest that access to capable AI models is no longer the only obstacle to enterprise adoption. Businesses must also ensure that the information feeding those systems is accurate, connected and relevant to the decisions they are expected to make.
According to the survey, 81% of businesses with fully AI-ready enterprise data report broad or strong returns from AI, compared with 12% of those whose data is only partially ready. That represents a 6.75-fold difference between the two groups, underscoring the association between data readiness and reported business outcomes.
The results do not establish that data readiness alone causes higher returns. However, they point to a practical challenge for organizations trying to move AI beyond isolated experiments: even capable models can struggle to deliver consistent value when business information is incomplete, fragmented or difficult to verify.
AI Investment Is Growing Faster Than Readiness
Nearly all respondents, 98%, report having active AI initiatives. Meanwhile, 65% expect to increase AI spending over the next 12 months, and 37% say they are scaling pilots into production, up from 34% in the previous quarter.
Yet the share reporting broad or strong AI returns remains at 28%, unchanged from Q3. Another 44% report pockets of ROI across some pilots, while only 9% say most pilots have scaled with clear gains.
The data-readiness figures help explain why expansion may not be translating into wider returns. Just 8% of businesses describe their enterprise data as fully ready for AI at scale. A further 33% say it is mostly ready, while 44% characterize it as partially ready.
The latest quarter showed some progress at the top: the share reporting fully ready data rose from 6% to 8%. But the proportion describing data as mostly ready fell from 36% to 33%. Most organizations therefore remain somewhere between early preparation and the level of readiness needed to support AI at scale.
For technology leaders, this creates a deployment challenge. Increasing model access or launching more pilots may not resolve problems involving inconsistent records, disconnected systems or insufficient business context.
The Difference Between Experimentation and Business Impact
Data readiness can affect how AI systems interpret an organization, its customers, suppliers and commercial relationships. In enterprise settings, a model may need to distinguish between businesses with similar names, understand how entities are connected or retrieve reliable information before recommending an action.
Without that context, AI-generated answers and decisions can be incomplete or inaccurate, even when the underlying model is capable. Poor-quality inputs can also increase the effort required to verify outputs and maintain reliable workflows.
Gary Kotovets, chief data and analytics officer at Dun & Bradstreet, said that verified business context helps distinguish AI pilots from initiatives that deliver measurable profit-and-loss impact. The company points to its D&B Commercial Graph as a source of connected business identity and relationship information that AI systems can use.
That positioning reflects Dun & Bradstreet’s commercial interest in business data, so its product claims should be evaluated separately from the survey’s broader findings. The survey itself suggests that enterprises may need to treat data preparation as a core part of AI deployment, rather than a secondary task after selecting a model.
For organizations, practical steps can include establishing data ownership, resolving duplicate or inconsistent records, connecting relevant sources and validating information used in AI workflows. They can also evaluate whether AI systems have access to the right context for a specific task and whether outputs can be checked against trusted business records.
Regional Gaps Highlight Uneven AI Preparedness
The survey also finds differences in AI readiness and returns across regions. Eastern Europe reports broad or strong AI ROI at 31%, followed by Asia Pacific and Western Europe at 30% each. These regions also report relatively high shares of mostly or fully AI-ready data, at 41% to 43%.
The Americas present a different picture. Approximately 68% of businesses there expect to increase AI spending, yet just 23% report broad or strong ROI, and 37% describe their data as mostly or fully ready. Africa reports the lowest figures among the regions cited for both broad or strong ROI, at 21%, and mostly or fully ready data, at 33%.
These comparisons show that greater spending intent does not necessarily correspond with stronger reported outcomes. They do not, however, establish that differences in data readiness fully explain regional performance; other factors may also influence adoption and returns.
Data Foundations Will Shape the Next Phase of AI
For enterprise AI teams, the central question is shifting from whether to invest in AI toward how to make deployments reliable, repeatable and economically useful.
The survey indicates that organizations with fully ready data are much more likely to report broad or strong ROI. But with only 8% of respondents reaching that level, data readiness remains a substantial constraint for many businesses pursuing AI at scale.
The next phase of adoption may therefore depend as much on data governance, integration and verification as on model capabilities. Enterprises that align their AI deployments with trustworthy information and measurable business objectives will be better equipped to determine whether systems are delivering value beyond isolated pilots.
Market Landscape
Enterprise AI adoption increasingly depends on the infrastructure surrounding the model, including:
- Data quality and governance: Ensuring that information is accurate, consistent and appropriately managed.
- Business identity and entity resolution: Connecting records to the correct companies, customers and suppliers.
- Data integration: Making information available across systems and workflows.
- AI evaluation: Measuring outcomes against business objectives rather than relying on pilot activity alone.
- Production readiness: Moving from limited experiments to repeatable deployments with measurable results.
Dun & Bradstreet’s survey highlights the gap between AI spending plans and data readiness. Because the results are survey-based, they reflect reported organizational experiences rather than independently measured financial performance or proof of causation.
Top Insights
- AI spending remains strong: 65% of businesses plan to increase AI investment over the next 12 months.
- Broad returns have stalled: 28% report broad or strong AI ROI, unchanged from the previous quarter.
- Data readiness is limited: Only 8% say their enterprise data is fully ready to support AI at scale.
- Reported ROI varies substantially: 81% of businesses with fully ready data report broad or strong ROI, compared with 12% with partially ready data.
- Production adoption is advancing: 37% report scaling AI pilots into production, up from 34% in Q3.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
