The Highest Opportunities for ROI Are Often Hidden
Every owner I talk to feels some version of the same pressure. Competitors are talking about AI. Customers are asking about it. Somebody at a conference described a result that sounds impressive. The worry that the business is falling behind is real.
The problem is how that anxiety affects people’s decisions. Anxiety makes us act, and the fastest way to adopt AI is to apply it to the first thing in the business that looks like a problem. Usually that is a visible, irritating, low-stakes task. Someone retypes data from one system into another, or a report takes three hours to assemble every Monday. It gets automated, it works, and the company concludes it is making progress on AI.
What is the impact? The first opportunity you notice is the one that is easiest to see, not the one with the most value behind it. Easy to see usually means small, contained and already tolerable. These are probably good things to fix. But if that clerical task was genuinely hindering the business, you would have fixed it years ago.
Meanwhile the work that would actually move the numbers stays untouched, because it is harder to look at. It crosses departments. It involves judgment. It has no single owner. Nobody points at it in a meeting because nobody thinks of it as one process.
This is part of why so much AI spending produces so little. A National Bureau of Economic Research survey of more than 6,000 senior executives in four countries found that roughly 90 percent reported no measurable productivity improvement from AI over the previous three years.¹ Those companies were not idle. They were busy on the wrong things.
There is a simple discipline that helps. Before approving an AI project, write down three candidates rather than one, and for each of them name the cost it removes or the revenue it enables, in dollars, over a year.
If you cannot achieve a meaningful ROI on any of the three projects, that is useful information. It does not mean the projects are irrelevant. But if you are willing to spend company time on AI adoption, these three may not be the right next project worth doing.
Sources
¹ National Bureau of Economic Research survey of more than 6,000 senior executives in the United States, United Kingdom, Germany and Australia, cited in “When Developing an AI Strategy, Beware the Urgency Trap,” Harvard Business Review, July 2026.