High Adoption, Low Absorption: Why AI Usage Metrics Mislead the Board
Why boards must look beyond AI usage metrics to measure workflow change, governance, and value
Boards are being shown more evidence of AI activity than evidence of AI transformation. Licences are being activated, pilots are being launched, employees are experimenting, and dashboards are filling up with usage data. Yet in many organisations, the operating model has barely changed.
In this article, we explore why AI adoption is not the same as AI absorption. The argument is simple: adoption shows that people can access and use AI; absorption shows that AI has changed how work is designed, governed, measured, and converted into value.
The adoption story looks better than the operating reality
The headline numbers make AI progress look impressive. McKinsey reported that 78% of organisations were using AI in at least one business function in 2024, while 71% said they were regularly using generative AI in at least one function. On the surface, this suggests AI has moved from experiment to mainstream business use.
But the value picture is much less convincing. McKinsey also found that more than 80% of respondents were not seeing tangible enterprise-level EBIT impact from generative AI, and only 21% said their organisations had fundamentally redesigned at least some workflows. That is the real issue. AI is being adopted faster than it is being absorbed.
This matters because boards can easily mistake movement for maturity. A growing number of users, pilots, prompts, and training completions can create the appearance of progress. But if workflows, handoffs, controls, decision rights, and value measures remain unchanged, the organisation may be busy with AI without becoming meaningfully better because of it.
Adoption is not absorption
AI adoption is evidence that the technology has entered the organisation. It includes licences, access, prompt volume, pilots, training sessions, and experimentation. These indicators are useful, especially in the early stages, but they mostly tell leaders whether people have started using AI.
AI absorption is different. It is evidence that AI has been incorporated into the operating model. It shows up when workflows are redesigned, decision latency improves, rework falls, exception handling changes, controls are embedded, and accountability for AI-assisted work becomes clear.
Adoption changes behaviour at the edge of the organisation. Absorption changes the operating model itself.
This distinction matters because individual productivity does not automatically become enterprise value. A person may finish a draft faster, but if the approval process is unchanged, the customer journey may not move faster. A team may generate more analysis, but if decision rights are unclear, the organisation may still make slow or poor decisions. AI absorption happens only when the surrounding system changes.
Why usage dashboards create false confidence
Many AI dashboards are built around the easiest things to measure: active users, licences, prompts, training completion, and pilot counts. These metrics are not useless, but they are incomplete. They show exposure to AI, not conversion into value.
The danger is that boards receive a picture of AI progress that looks precise but is strategically shallow. A dashboard showing rising prompt volume may look positive, even while approval bottlenecks, rework, and customer response times remain unchanged.
High usage may include casual experimentation, duplicated work, low-value prompting, or employees using AI outside approved processes. It may say very little about whether AI has improved cycle time, quality, cost to serve, risk posture, or customer outcomes.
This is why the board question needs to change. Instead of asking, “How many people are using AI?”, leaders should ask, “Where has AI changed the way work gets done, and what measurable outcome has improved as a result?”
Why local productivity does not always become enterprise value
One of the biggest traps in AI reporting is the assumption that local time savings automatically convert into business value. They often do not. Time saved at task level can be absorbed by review, rework, coordination, extra output demands, or bottlenecks elsewhere in the process.
This is the difference between task productivity and system productivity. AI may help an employee write faster, summarise faster, or analyse faster. But if the workflow still depends on slow approvals, unclear ownership, fragmented data, or manual exception handling, the end-to-end process may barely improve.
That is why workflow redesign matters so much. McKinsey’s research points to workflow redesign as one of the strongest factors associated with bottom-line impact from generative AI. The value is not created simply because people use AI. It is created when work is rebuilt around what AI changes.
The hidden risk of high adoption
High adoption can also create hidden risk. When employees use AI without approved tools, clear policies, or proper controls, the organisation may gain speed while losing visibility. Gartner has warned that shadow AI is already widespread, with many organisations suspecting or having evidence of employees using prohibited public GenAI tools.
This creates a governance problem. Sensitive data may be pasted into unapproved systems. AI-generated outputs may be used in client work without review. Decisions may be influenced by tools that are not monitored, logged, or governed. In that environment, adoption can rise faster than accountability.
The issue is not that employees are wrong to experiment. Often, they are responding to real workflow friction. The leadership failure is allowing official systems, guidance, and governance to lag behind actual behaviour.
What absorption looks like in practice
A genuinely absorbed AI use case has a different profile. It has a named workflow, a business owner, a baseline, a measurable outcome, a control model, and a clear route to scale. It does not sit as a disconnected pilot or personal productivity hack.
Absorption becomes visible when workflows are redesigned, approvals change, review burden falls, cycle time improves, ownership becomes clearer, and controls are embedded into how work actually runs.
For example, in customer service, absorption might mean AI is embedded into triage, response drafting, escalation routing, and quality review. In finance, it might mean AI supports forecasting or controls testing with clear data lineage and human approval points. In HR, it might mean AI assists with workforce planning or employee queries while fairness, privacy, and escalation rules are built into the process.
Absorption is therefore not just “AI is being used.” It is “AI has changed the workflow, and we can prove the change improved performance without weakening control.”
What boards should measure instead
Boards should still track adoption, but adoption should sit inside a wider absorption dashboard. The goal is to understand whether AI is changing the business, not just whether people are touching the tools.
A better board dashboard should include:
Adoption depth: who is using approved AI tools, in which roles, and in which workflows
Workflow absorption: what share of a target process has actually changed because of AI
Cycle time: whether work is moving faster from trigger to completion
Quality and rework: whether outputs are improving or creating more correction effort
Decision latency: whether AI is helping decisions happen faster and better
Value realised: cost reduction, revenue uplift, margin improvement, or productivity gains where measurable
Risk and control coverage: whether use cases have owners, logs, review points, and incident routes
Shadow AI exposure: where unauthorised tools or unmanaged workflows are still being used
Workforce impact: whether AI is improving work or simply increasing intensity
Scaling rate: how many use cases have moved from experiment to governed production
This moves the conversation from activity to conversion. It also gives boards a more honest view of where AI is working, where it is stalled, and where risk is building.
The leadership shift now required
The next phase of AI leadership is not about proving that people can use AI. That phase is already underway. The harder task is proving that AI is being absorbed into the organisation in a way that changes performance.
That requires leaders to focus capital on fewer, better-defined workflows. It requires owners who are accountable for outcomes, not just experimentation. It requires governance that is embedded into the workflow, not written as a separate policy after deployment. And it requires boards to challenge dashboards that show activity without showing value.
The organisations that succeed will not be the ones with the highest number of pilots or the loudest AI narrative. They will be the ones that can show, with evidence, that AI has changed how work gets done and that those changes are producing measurable, governed, sustainable value.
Conclusion
AI adoption is real, but absorption is still scarce. Boards should be careful not to mistake licences, pilots, prompts, and training completions for transformation. The real test is whether AI has changed workflows, decisions, controls, accountability, and outcomes. The board’s job is therefore to ask better questions: not just “who is using AI?”, but “where has AI been absorbed into the operating model, and what value can we prove?”
The organisations that outperform in the AI era will not be those with the most visible adoption. They will be the ones that redesign their operating model deeply enough for AI to become part of how the business actually runs.
End.
Read more articles from Kieran Gilmurray here
Listen to the audio version of this article on my Spotify channel
References
McKinsey & Company, The State of AI: How Organizations Are Rewiring to Capture Value
https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-how-organizations-are-rewiring-to-capture-valueBoston Consulting Group, AI at Work 2025: Momentum Builds, but Gaps Remain
https://www.bcg.com/publications/2025/ai-at-work-momentum-builds-but-gaps-remainDeloitte, The State of Generative AI in the Enterprise: Q4 Report
https://www.deloitte.com/content/dam/assets-shared/docs/about/2025/quarter-4.pdfGartner, Gartner Survey Finds Regular AI System Assessments Triple the Likelihood of High GenAI Value
https://www.gartner.com/en/newsroom/press-releases/2025-11-04-gartner-survey-finds-regular-ai-system-assessments-triple-the-likelihood-of-high-genai-valueGartner, Gartner Identifies Critical GenAI Blind Spots That CIOs Must Urgently Address
https://www.gartner.com/en/newsroom/press-releases/2025-11-19-gartner-identifies-critical-genai-blind-spots-that-cios-must-urgently-address0IBM, 2025 CEO Study: 5 Mindshifts to Supercharge Business Growth
https://newsroom.ibm.com/2025-05-06-ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdlesForrester, Three Years Into GenAI, Enterprises Are Still Chasing Its True Transformative Value
https://www.forrester.com/press-newsroom/forrester-three-years-into-genai-enterprises-are-still-chasing-its-true-transformative-value/MIT Sloan School of Management, How AI Is Reshaping Workflows and Redefining Jobs
https://mitsloan.mit.edu/ideas-made-to-matter/how-ai-reshaping-workflows-and-redefining-jobsStanford HAI, The 2025 AI Index Report
https://hai.stanford.edu/ai-index/2025-ai-index-reportNIST, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence-profile