The pattern is becoming familiar across European enterprises. Organisations implement AI tools, see initial productivity jumps, then watch the gains flatten out. Executives find themselves asking difficult questions about whether the investment has actually paid off. The problem, according to experts, isn't a shortage of solutions—it's disagreement about what needs fixing. Some point to governance gaps, others to deployment issues. But the real obstacle may be simpler: most organisations haven't decided what success actually means, and they're not measuring the right things.
Learning and development professionals are caught in the middle of this puzzle. These are the people best positioned to build the skills organisations need for AI adoption, yet they face the same measurement crisis affecting the rest of their companies. A survey of more than 800 workplace learning professionals across more than 25 countries offers insight into how to move from simply adopting AI to actually getting returns from it.
When speed becomes the only metric, results suffer
How learning teams deploy AI depends largely on what their organisation values and tracks. When companies reward higher token usage without tying it to specific outcomes, teams optimise for volume. The issue emerges when this metric-chasing happens in isolation, disconnected from clear business goals. Organisations need to articulate the purpose: we want more AI use to achieve X. Without that clarity, productivity gains don't translate into meaningful performance improvement.
The survey data illustrates this disconnect sharply. Three-quarters of respondents identified faster content creation as the clearest benefit they've observed from AI adoption—essentially, the ability to produce training materials in less time and at greater volume. Yet only 11% reported seeing improved on-the-job performance, which is what most workplace training is supposed to deliver in the first place. The measurement gap is even starker: 72% of learning teams consistently track participation and completion rates, but only 35% consistently measure whether training actually changes how people perform their jobs.
This pattern suggests either that leaders are rewarding the wrong metrics, or that organisations haven't collectively agreed on what winning looks like. The survey responses lean toward the latter explanation. More than one in five respondents identified the absence of agreed success metrics as their primary barrier to demonstrating impact.
Mandates without permission, tools without data
Measurement challenges aren't the only structural problem. Governance and access constraints create additional friction. When leadership mandates AI adoption, teams often still need approval from their department or across functions just to test a new tool. Some organisations restrict teams to a pre-approved list of AI platforms, eliminating the ability to experiment beyond those boundaries.
Data access presents another barrier. Most learning teams believe AI is driving improvements, but struggle to produce evidence. When asked about the biggest obstacle to proving impact, 53% cited organisational constraints—including both the lack of agreed success metrics and the inability to access data or measurement tools. The problem frequently stems from data fragmentation: the information needed to assess AI's impact lives in separate systems—a learning management platform here, an HR information system there, survey tools elsewhere, and AI video platforms in yet another location.
If learning teams are using AI primarily to generate training content faster without evaluating whether that content changes behaviour, they're essentially running an automated content factory. Until organisations define outcomes beyond speed, that's all they'll have to show for their AI investment.
Three structural shifts that unlock ROI
The encouraging finding: roughly a third of learning teams report genuine performance gains from AI—faster time to proficiency, improved on-the-job performance. These teams share common traits: freedom to pilot or build AI solutions, and the ability to use AI to evaluate their own work. Many of these advantages depend on organisational decisions beyond learning's control.
Based on the survey data, three structural changes can help leaders align their AI strategy with actual returns.
- Align on purpose first, then on how usage gets rewarded and measured. Without a defined outcome, teams default to what's easiest to demonstrate—in learning, that's faster content production. Instead, learning leaders should work with senior management to identify the performance changes they want to see and how to measure them, before building any new training. Yes, that includes AI enablement training.
- Invest in data integration before scaling AI solutions or automating additional workflows. Ensure teams across the organisation can access the data they need to measure AI's actual impact.
- Bring learning into your AI strategy as a partner. If the goal is automating manual work, the secondary goal should be freeing up time for strategic capability building—teaching people how AI changes the way work gets done. Learning teams are uniquely positioned to lead this effort because they can identify where training fits the solution and where it doesn't.
Only when these structures are in place can training actually help organisations close the gap between adopting AI and realising returns from it.
Source: The Next Web
