You Are Probably Measuring the Wrong Things
Somewhere in your organisation, there is a dashboard tracking AI adoption. It shows prompt volume, tool logins, the number of departments with active pilots, and maybe a usage trend line that goes up and to the right. Leadership reviews it quarterly. Everyone agrees it looks healthy.
The problem is that none of those numbers tell you whether AI is actually changing anything.
Tool usage is an activity metric. It tells you that people are interacting with the technology. It says nothing about whether the workforce is gaining new capabilities, whether roles are evolving to take advantage of what AI enables, or whether business performance is improving as a result.
McKinsey's 2025 State of AI survey found that while 88% of organisations use AI in some capacity, only 39% can point to measurable financial returns. The gap between those two numbers is largely an outcomes gap. Organisations are tracking adoption, not impact.
The enterprises that are closing that gap have moved their measurement frameworks around three specific outcomes: skills gained, jobs transformed, and performance improved. Everything else is supporting data.
Outcome 1: Skills Gained
The first outcome that matters is whether the workforce is actually developing new capabilities as a result of the AI programme.
This sounds obvious, but most organisations do not track it with any rigour. They run AI training programmes and measure completion rates. They deploy tools and measure logins. What they do not measure is whether employees can do something today that they could not do six months ago, and whether that new capability connects to something the business needs.
Tracking skills gained requires a few things to be in place. The organisation needs a coherent skills taxonomy, so that capability development can be defined and measured consistently. It needs assessment mechanisms that go beyond course completion, because finishing a module does not prove competence. And it needs a feedback loop that connects skills data to business outcomes, so that leadership can see which capabilities are actually driving value.
According to Deloitte's 2025 Global Human Capital Trends report, organisations that adopt skills-based approaches to workforce development significantly outperform those that rely on traditional role-based models. The AI programme should be accelerating this shift. If the skills data is not improving, the programme may be generating activity without building capability.
Outcome 2: Jobs Transformed
The second outcome is whether jobs across the organisation are actually changing shape.
AI adoption should, over time, reshape how work gets done. Tasks that were manual become automated. Decisions that relied on intuition get supported by data. Processes that took days compress into hours. When these changes add up across enough tasks, the job itself transforms. Employees spend less time on low-value activities and more on the work that requires judgement, creativity, and relationship-building.
That transformation should be visible and measurable. Are specific roles spending materially less time on administrative tasks than they were a year ago? Have new responsibilities been added to existing roles that reflect AI-augmented capabilities? Are teams operating with higher throughput without proportional headcount increases?
If the AI programme has been running for a year and no job in the organisation looks meaningfully different, that is a signal worth investigating. It usually means the technology is being used for convenience rather than for transformation, bolted onto existing processes without redesigning those processes to take advantage of what AI makes possible.
The World Economic Forum's Future of Jobs Report 2025 projects significant restructuring of roles globally by 2030. Organisations that track job transformation as an explicit outcome of their AI strategy will be better prepared for that restructuring. Those that treat roles as static will find themselves scrambling to catch up.
Outcome 3: Performance Improved
The third outcome is the one the board cares about most. Is business performance improving?
This is where the measurement challenge gets genuinely difficult, because business performance is influenced by many factors, and isolating the contribution of the AI programme requires discipline.
The strongest approach is to define performance indicators at the outset and track them consistently. Those indicators should be specific to the business units where AI is being deployed. For a sales organisation, that might be pipeline velocity or win rates. For an operations function, it might be throughput or error rates. For a customer service team, it might be resolution time or satisfaction scores.
The AI programme should be able to draw a credible line from tools deployed, to skills developed, to jobs reshaped, to performance moved. Each link in that chain is measurable. When all four are tracked together, the organisation can tell a clear story about whether its AI investment is producing returns.
PwC's 2026 AI predictions report reinforces this: the organisations demonstrating measurable AI returns are the ones that concentrate their resources on high-impact applications and track outcomes rather than activity. Broad tool deployment without outcome measurement produces impressive dashboards and disappointing board conversations.
Why These Three Outcomes Belong Together
Skills, jobs, and performance are connected in a sequence that matters.
Skills gained without jobs transformed means the workforce is learning things it has no opportunity to apply. That leads to frustration and wasted investment.
Jobs transformed without skills gained means the organisation is redesigning roles for a workforce that cannot fill them. That leads to operational risk and hiring pressure.
Both without performance improved means the programme is generating internal change with no external impact. That leads to budget questions.
When all three are tracked together, they create a coherent narrative. The workforce is developing new capabilities. Those capabilities are reshaping how work gets done. And the reshaped work is producing better business results. That narrative is what sustains executive investment through the multi-year commitment that real AI transformation requires.
Reframing the Dashboard
The activity metrics are not useless. They are just insufficient. Prompt volume and tool adoption have their place as leading indicators. But they should not be the headline.
If your organisation is ready to move its AI measurement framework from activity tracking to outcome tracking, we help enterprises build that discipline into their programmes from the start.
Talk to our team at kydongrp.com/contact
Sources: McKinsey & Company. "The State of AI in 2025." https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai Deloitte. "2025 Global Human Capital Trends." https://www2.deloitte.com/us/en/insights/focus/human-capital-trends.html World Economic Forum. "Future of Jobs Report 2025." https://www.weforum.org/publications/the-future-of-jobs-report-2025/ PwC. "2026 AI Business Predictions." https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-predictions.html