AI is increasingly delivering measurable business results. New global research from Dun & Bradstreet, conducted among 10,000 companies across 32 countries, shows that 60% of organizations are now achieving at least some measurable return on investment (ROI) from AI. At the same time, only 24% report broad or significant results, while 56% plan to increase their AI investments.
These findings suggest that AI is entering a new phase. The focus is shifting from experimenting with possibilities to identifying where AI genuinely creates value. In this article, we explore what the D&B research reveals about AI ROI, why scaling requires deliberate choices, and the role data plays as AI becomes embedded in business processes.
AI Is moving from experiment to business process
For many organizations, AI is no longer limited to a pilot project or a single-team initiative. The research shows that 97% of organizations have active AI initiatives. In addition, 30% are now scaling AI into production environments, while 26% are operationally deploying AI across multiple core business processes.
As a result, the role of AI is changing. What once served primarily as a tool for exploring new possibilities is increasingly becoming part of processes where information is handled daily and decisions are prepared. Examples include analyzing customer developments, gathering supplier intelligence, or identifying changes that require attention.
At this stage, technical functionality alone is no longer enough. AI must fit seamlessly into the processes where employees use it. A fast result only becomes valuable when users understand what it means and how they should act on it.
Interesting read: Agentic AI: from hype to practical reality
Not every successful AI application needs to be scaled up
The fact that organizations are already seeing returns from AI is an important development. However, the gap between the 60% reporting some measurable ROI and the 24% achieving broad or significant results highlights that success does not automatically translate across the organization.
An application that helps employees process documents more efficiently may deliver immediate time savings. However, AI systems used for customer selection, supplier monitoring, or risk assessment require a different level of reliability. In these cases, outcomes must not only be delivered quickly but also be accurate, relevant, and trustworthy enough to support decision-making.
As organizations increase their AI investments, it becomes increasingly important to define for each application:
- Which process it improves.
- What outcome actually creates value.
- What information is needed to evaluate that outcome.
- When human oversight remains necessary
This helps prevent organizations from scaling AI simply because an initial use case performed well, without understanding whether the same value can be achieved in other processes.
Scaling reveals the real challenges
Now that AI is increasingly being deployed in production, the conditions for successful implementation are also becoming clearer. Only 5% of organisations indicate in the D&B study that the own data is fully ready for AI.
Organizations identified several barriers to further AI adoption:
- 50% cite limited access to data.
- 44% sees privacy and compliance risks as an obstacle.
- 40% point to challenges related to data quality and data integrity.
- 38% experience insufficient integration between systems.
- 37% report a shortage of professionals with the necessary AI skills.
In addition, only 10% are highly confident in their ability to identify and mitigate AI-related risks.
These findings make it clear that scaling AI is not just about the technology itself. When AI is applied to processes involving customers, suppliers, and business relationships, employees need to understand how insights are generated and whether the underlying information is current, reliable, and actionable.
For example, a signal about a customer is only valuable when it is clear which organisation it relates to. An analysis of a supplier only becomes truly useful when relevant changes and relationships are visible. In processes like these, the quality of the available data determines whether AI produces an interesting signal or an insight that employees can actually act on.
Interesting read: Everyone wants to do something with AI, but is your organization ready for it?
From AI experiment to lasting value
The first measurable AI results are already visible. For organizations, this marks the beginning of the next phase: moving beyond discovering what AI can do and focusing on where it can deliver lasting business value.
Organizations that embed AI into business processes will therefore look beyond speed and efficiency alone. The combination of a well-defined use case, clear governance, and reliable data ultimately determines whether an initial success can evolve into sustainable value over the long term.
Want to learn more about how business information can support AI applications within your organization? Get in touch with our experts.
