Over the last two years, the enterprise conversation around AI has shifted. AI is embedded across functions, workflows, and teams. Pilots are running, copilots are deployed, and experimentation is widespread.
And yet, something isn't adding up.
Despite this momentum, most organisations still struggle to clearly articulate the value they are getting from AI. Investments continue to grow, but confidence in returns remains low. The issue is no longer adoption, it is clarity.
At first glance, many organisations appear to be making strong progress. Teams have access to tools. Use cases are expanding. Dashboards show growing activity. But when leaders are asked a simple question: what measurable business impact has AI delivered?
The answers are often incomplete.
This gap stems from three underlying issues.
Lack of Attribution
AI activity is visible, but its contribution to outcomes is not. It is relatively easy to track how many tools are deployed or how many employees are using AI. It is far harder to quantify how those tools are improving revenue, reducing costs, or mitigating risk. As a result, many AI initiatives are justified on potential rather than proof.
Persistence of Pilots
Organisations have learned to launch AI initiatives quickly, but not necessarily to scale them effectively. This pattern is not new. Previous technology waves followed a similar trajectory: early enthusiasm, fragmented experimentation, and only a small percentage of initiatives translating into enterprise-wide value. AI is at risk of repeating this cycle.
Depth of Adoption
Access to AI does not guarantee meaningful usage. In many cases, AI is present in workflows but not deeply integrated into how work gets done. Employees may experiment with tools, but without structured guidance, role-specific enablement, and clear use cases, adoption remains shallow.
Relative ease of visibility
Tracking AI activity vs. measuring business value
Together, these factors create the illusion of progress. Organisations appear to be advancing, but the link between activity and value remains weak.
Compounding this challenge is the nature of AI itself. Unlike traditional technology investments, AI does not produce a clean before-and-after effect. Its capabilities evolve continuously, and so does the organisation's ability to use it. The baseline keeps shifting. Improvements are gradual, distributed, and often indirect.
This makes traditional ROI models insufficient. AI does not deliver value in a single moment. It accumulates over time, as adoption deepens and workflows adapt.
The implication is straightforward but important.
The challenge is no longer deploying AI. It is understanding, measuring, and scaling its impact.
Organisations that continue to focus only on adoption will struggle to justify ongoing investment. Those that focus on measurement and value realisation will be able to scale with confidence.