Business 6 min read By Arthur Ellington
McKinsey partner warns AI's easiest wins are misleading CEOs
McKinsey research shows AI adoption has stalled at 89% while only 6% of companies report meaningful earnings impact. A senior partner argues that leaders are mistaking isolated productivity gains in contact centres and software teams for a universal playbook, and that genuine returns require redesigning work around clear outcomes rather than simply inserting AI into existing processes.
Artificial intelligence adoption across global organisations has barely shifted in the past year, yet the share of companies attributing meaningful earnings to the technology remains stubbornly low, according to new research from McKinsey. The consultancy found that 89 per cent of organisations used AI in at least one business function, up from 88 per cent a year earlier. High performers, defined as those attributing at least 5 per cent of earnings before interest and taxes to AI, held flat at 6 per cent in both years. Thirty-seven per cent of companies reported some positive effect on earnings, but the gap between broad use and genuine financial payoff is not closing.
Writing for Fortune, a McKinsey senior partner argues that executives are being misled by the clearest early successes of AI. A study of around 5,200 customer-support agents found that AI assistance increased the number of issues resolved per hour by 15 per cent. Randomised trials involving roughly 4,900 software developers found that those given an AI coding assistant completed about 26 per cent more tasks. These are meaningful gains, but they do not establish that adding AI to any business process will improve a company's earnings.
Contact centres and software teams had crucial advantages before today's AI models arrived. Contact centres had spent decades organising high volumes of work into queues, tracking outcomes and building a body of past interactions. Software teams had developed testing, continuous integration and code review practices that made it possible to inspect and correct new work. AI could enter those settings and make an existing system faster. That is valuable, but it is a different challenge from redesigning work that has never been organised around a clear outcome or a way to check whether it was achieved.
The partner offers the example of a bank using AI to read documents for small-business loans. Faster document review might save two days, but if an application still waits for separate handoffs among sales, credit, compliance and operations, the customer may see little improvement. A faster first step does not resolve the delays and errors that arise later. Redesign would start with the decision the bank needs to make: what evidence is required to approve a sound loan, who has the authority to make that decision, and which exceptions need specialist review. AI could help gather and check the evidence, while a named team owns the application from submission to decision. The bank would measure time to decision, error rates and the loans it can responsibly serve, rather than counting only hours saved reading files.
That shift requires choices about authority, risk and what employees will do with the time they regain. Software cannot make those choices for a leadership team. The partner describes watching executives nod at AI demonstrations without asking what would change after the demonstration ends: who owns the outcome when a process crosses several functions, who can remove a handoff, and how anyone will know that faster work produced a better result. Without answers, a promising pilot can remain a pilot. Employees notice that uncertainty too, asking whether they are training their replacement. Managers cannot offer much reassurance if leadership has described only the tasks AI might perform, not the work people will do next.
McKinsey research published in July offers a reason to ask harder questions. Among leaders reporting on organisations in the earliest stage of AI adoption, those whose workflows had been redesigned were 5.3 times as likely to report enterprise-level value as those whose workflows had not: 32 per cent versus 6 per cent. That finding should change the first question a chief executive asks about AI transformation. Before choosing a tool, decide what outcome the company needs and which parts of the work no longer serve it. Give one leader responsibility for the redesigned process, establish how its decisions will be checked, and decide whether the capacity it frees will support growth, better service or lower cost.
The gains in contact centres and coding are genuine, but they are a starting point, not a template that can be dropped into any company. The larger opportunity belongs to leaders willing to change the work around the technology. The partner also acknowledges that management consulting bears some responsibility, since consultants often earn their keep by removing waste from an existing process, a practice known as optimisation. Firms are now beginning to rethink the purpose of accumulated organisational processes and the value trade-offs they embody. For chief executives, the message is that AI adoption alone is not a strategy. The returns will come from deciding what the business is trying to achieve, who owns the outcome, and how performance will be measured once the technology is in place.
4
