AI investment is accelerating. Employees are saving time. Productivity is improving. But how much of that activity is translating into measurable business performance?
New research from McKinsey suggests that the next challenge isn't simply getting more people to use AI. It's fundamentally rethinking how organisations work because AI is making people more productive.
Developers are writing code faster. Analysts are processing information in minutes rather than hours or even days. Marketing teams generate content, customer service agents resolve enquiries, and managers use AI to support everything from research to decision-making.
Across organisations, employees are discovering new ways to work more efficiently.
And businesses are investing accordingly.
According to Boston Consulting Group's September 2026 Applied AI Index, corporate AI spending has almost doubled in a year, rising from approximately 1.7% to 3.3% of revenue.
That's a substantial investment.
But here's the question.
If employees are becoming more productive, are the organisations employing them becoming more valuable?
And can anyone demonstrate where that value is appearing?
AI Adoption Is Outpacing Enterprise Value
Yesterday (October 13), McKinsey published The AI Operating System: A New Way for CEOs to Run the Enterprise.
One finding particularly caught our attention.
According to McKinsey's 2026 State of AI research, 40% of respondents at large companies report scaling AI agents in at least one function.
Yet only 6% attribute an improvement of at least 5% in EBIT to AI.
These figures measure different things, so they don't translate directly into a conversion rate from adoption to financial return. But they do illustrate an important gap between deploying AI and generating substantial enterprise-level value.
Other research tells a similar story.
IBM's 2025 CEO study found that executives believed only 25% of AI initiatives had delivered their expected ROI, while just 16% had scaled across the enterprise.
Deloitte's October 2025 research, covering 1,854 executives across Europe and the Middle East, found that 85% of organisations had increased AI investment, yet only 6% reported satisfactory payback on a typical AI use case within 12 months.
But there's encouraging evidence too.
BCG's September 2026 research reports that nearly half of companies are now generating meaningful value from AI.
The definitions of value differ across these studies. Meaningful operational benefits aren't necessarily the same as verified financial returns.
Nevertheless, the picture is becoming clearer.
AI is creating value. But adoption, productivity and financial performance are three different things.
The challenge is connecting them.
That becomes particularly difficult when AI investment lacks clear leadership alignment and accountability
The £6 Million Productivity Illusion
Let's imagine a business with 1,000 employees.
Through AI adoption, each employee saves an average of three hours a week.
That's 3,000 hours of additional capacity every week.
Or approximately 150,000 hours annually, assuming 50 working weeks.
Apply an illustrative employment cost of £40 per hour, and the theoretical capacity released is worth £6 million.
An impressive figure for the board presentation.
But has the organisation actually saved £6 million?
Not necessarily.
Unless that capacity is converted into something valuable, the financial benefit may be limited.
Perhaps employees produce more work. Perhaps they spend more time in meetings. Perhaps they have more time to focus on customers, solve difficult problems or develop new ideas.
Perhaps they're simply less stressed.
That last outcome shouldn't be dismissed. Better employee experience matters.
But none of these outcomes automatically translate into £6 million of financial benefit.
Time saved is not automatically value created.
And organisations risk confusing the two when they measure AI success primarily through access, adoption, usage and self-reported hours saved.
These are useful indicators, but they don't tell us what has actually improved for the business.
Productivity Gains Are Real. Capturing Them Is Harder.
This isn't an argument against AI productivity.
There is increasingly strong evidence that the benefits are real.
In a study of more than 5,000 customer support agents, researchers Erik Brynjolfsson, Danielle Li and Lindsey Raymond found that a generative AI assistant increased customer issues resolved per hour by approximately 14%.
Less experienced and lower-skilled employees saw particularly significant improvements, averaging around 34%.
The researchers also identified improvements in customer sentiment and employee retention.
That's meaningful operational progress.
But even here, higher productivity doesn't automatically tell us whether the organisation has achieved a positive financial return.
That depends on implementation costs, demand, service quality and what happens to the additional capacity.
The technology creates an opportunity.
The organisation still has to capture it.
And this is where McKinsey's latest thinking becomes particularly interesting.
What If the Problem Isn't Employee Productivity, But How We Organise Work?
McKinsey introduces the concept of an AI operating system.
Not another application sitting on an employee's desktop.
Rather, a connected layer of enterprise intelligence that draws on an organisation's data, institutional knowledge, business rules and expertise to help people make decisions and act on them.
The ambition is to connect information, decisions and execution across the organisation.
And that challenges something much more fundamental than individual productivity.
How we manage businesses.
Consider a familiar example.
An employee uses AI to prepare a monthly management report.
Previously, it took two days.
Now it takes two hours.
Excellent.
But perhaps the report still spends another week being reviewed, approved and circulated.
Perhaps the next management meeting isn't until the following month.
Perhaps nobody acts on the information anyway.
The employee has become dramatically more efficient.
The organisation hasn't necessarily become more effective.
Now imagine a different approach.
Instead of producing a monthly report, the organisation continuously monitors the information that matters.
Exceptions are identified automatically.
Managers are alerted when intervention is needed.
Decisions happen closer to the point where problems or opportunities emerge.
Suddenly, we're not talking about producing the same report faster.
We're questioning whether the report — and the management process surrounding it — needs to exist in its current form.
This is the distinction McKinsey is exploring.
And it's one I think deserves far more attention.
The opportunity isn't simply to work faster. It's to work differently.
The Real Opportunity Crosses Organisational Boundaries
One of the most compelling arguments in McKinsey's article is that significant value often comes from redesigning workflows that cross functions.
That's important because organisations are full of boundaries.
Technology and business.
Finance and operations.
Sales and delivery.
Risk and compliance.
Each function has its own systems, processes, priorities and performance measures.
This fragmentation reflects a familiar disconnect between business priorities and technology execution
Making one team more productive doesn't necessarily improve the overall outcome.
Sometimes it simply moves the bottleneck somewhere else.
Consider customer onboarding in a bank.
AI might help an employee review documents or identify missing information more quickly.
Useful.
But the bigger opportunity is to examine the entire customer journey.
Why does information move between multiple teams? Where do approvals get delayed? Which checks are duplicated? What creates unnecessary work for customers?
The business value would come from outcomes such as faster onboarding, fewer errors, lower operating costs, better customer experience and stronger controls.
That requires more than an AI tool.
It requires technology, operations, risk, finance and business leaders to redesign the process together.
McKinsey offers a manufacturing example where an AI-enabled redesign of planning, routing and pricing workflows created an estimated annual value opportunity of $140 million to $180 million.
That's an estimate of potential value, rather than a claim that the entire amount has already been realised.
But it illustrates the scale of the opportunity when AI is applied to end-to-end business performance rather than isolated tasks.
And it changes the conversation from productivity to economics.
AI Could Change Management Itself
Perhaps the most provocative part of McKinsey's report isn't about technology at all.
It's about management.
Many of our established management practices exist because information has historically been difficult to gather, interpret and distribute.
Monthly reporting.
Quarterly reviews.
Multiple layers of approval.
Information travelling up through management before decisions travel back down.
AI could remove some of those constraints.
Many of these practices reflect leadership structures designed around caution and control.
McKinsey suggests that organisations may need to move from periodic reviews towards continuous monitoring and intervention, from information-sharing towards judgment, and from hierarchical decision-making towards more distributed responsibility.
That's a significant shift.
But it also raises difficult questions.
If information is continuously available, which meetings are still necessary?
If AI can analyse options and recommend actions, which decisions genuinely require senior management?
If employees closer to customers have better information, why should every decision travel through multiple layers?
And if agents can execute parts of a process, who remains accountable when something goes wrong?
These aren't simply technology questions.
They're questions about leadership, governance, organisational design and trust.
AI may make existing management practices more efficient. But its greater impact could come from making some of them unnecessary.
The Missing Link is Organisational Capability
This is where our work with organisations on digital transformation keeps bringing us back to the same challenge.
Technology is rarely the whole problem.
An organisation might have excellent engineers, sophisticated AI tools and substantial investment.
But if business leaders can't identify valuable opportunities, if functions struggle to collaborate, or if nobody is prepared to challenge established ways of working, much of that investment may fail to deliver its potential.
McKinsey's proposed AI operating system is an interesting technical and organisational response.
But an operating system doesn't redesign an organisation by itself.
People still need to decide what matters.
They need to understand the economics.
They need to challenge assumptions, manage risks, redesign responsibilities and bring colleagues alongside.
That also means understanding how differently people experience organisational change
And those capabilities need to exist across the business.
Technology leaders need a stronger understanding of strategy, finance and commercial outcomes.
Business leaders need enough understanding of AI and technology to participate confidently in decisions.
Finance leaders need credible ways to evaluate value.
HR and L&D leaders need to help people develop the skills and behaviours required for different ways of working, and make those changes last after the programme ends.
The traditional separation between technology and business becomes increasingly difficult to justify.
Often, what looks like a technology problem is really a shared organisational capability gap.
The next phase of AI transformation is as much about developing organisational capability as deploying technical capability.
Show Me The Money
There is another important dimension to this conversation.
AI itself has an economic cost.
Models, infrastructure, integration, security, governance, maintenance and human oversight all require investment.
McKinsey notes that even as token costs have fallen dramatically, overall AI spending at many organisations has continued to rise.
So the business case cannot stop at how many hours an AI tool saves.
It needs to consider the full economics of the process.
A faster process might still be more expensive if it requires costly infrastructure or extensive human review.
A cheaper automated decision might create unacceptable risk.
And an impressive productivity improvement might have limited financial value if the organisation doesn't change how it deploys its people and resources.
This suggests five useful dimensions for measuring AI value:
- Cost: Are we reducing the total cost of delivering an outcome?
- Revenue: Are we creating new opportunities or improving commercial performance?
- Customer experience: Are we improving speed, quality or satisfaction?
- Risk: Are we reducing errors, improving controls or strengthening decisions?
- Capacity: Are we enabling people to do more valuable work?
Not every benefit needs to appear immediately as additional profit.
But there should be a credible connection between the investment, the operational improvement and the business outcome.
Otherwise, we're measuring activity rather than value.
A Different Conversation In the Boardroom
McKinsey's latest report makes a compelling case for rethinking the enterprise around AI.
I'm not convinced every organisation needs to adopt the specific AI operating system architecture it describes.
The right approach will depend on the business, its technology estate, regulatory obligations, maturity and strategic priorities.
But the underlying challenge is difficult to dispute.
We've spent considerable energy asking how AI can make employees more productive.
Perhaps we now need to spend more time asking how AI can make organisations more effective.
That means changing the questions we ask.
Not simply:
How many employees are using AI?
How many agents have we deployed?
How many hours are we saving?
But:
What business outcomes are improving because of AI?
Where are we capturing the value of the capacity being released?
Which processes have we fundamentally changed rather than merely accelerated?
Which management practices are no longer necessary?
And can we demonstrate that the benefits justify the investment?
Because an organisation can have thousands of employees using AI every day without fundamentally improving its financial performance.
The next phase of AI transformation won't be defined simply by how quickly organisations deploy the technology.
It will be defined by how effectively they connect individual productivity with organisational capability, better decisions and measurable business outcomes.
AI is making employees more productive.
Now organisations need to demonstrate what that's worth.
Turn AI Productivity Into Business Performance
AI investment creates opportunities. Realising their value requires leaders who can connect technology, organisational capability and commercial outcomes.
CTO Academy helps organisations develop that capability across technology and business leadership teams.
Explore our Corporate Leadership Programmes →
Research referenced
- McKinsey (13 October 2026): The AI Operating System: A New Way for CEOs to Run the Enterprise
- Boston Consulting Group (September 2026): Applied AI Index 2026 — AI Is Starting to Pay Off
- Deloitte (October 2025): AI ROI: The Paradox of Rising Investment and Elusive Returns
- IBM Institute for Business Value (May 2025): CEOs Double Down on AI While Navigating Enterprise Hurdles
- Brynjolfsson, Li and Raymond: Generative AI at Work — NBER Working Paper 31161