I was reading about ROI measurement. The stat that stopped me: only 31 percent of business leaders think they’ll be able to evaluate AI ROI within six months.
That’s it. Three out of ten.
Which means seven out of ten companies are building AI initiatives with zero visibility on whether it’s working. Not a long timeline. Not “we’ll measure in two years.” Right now. Today. They don’t know.
Why is this happening?
The problem is that AI benefits are weird. They’re indirect. They’re long-tail.
Like, let’s say you implement an AI chatbot for customer support. How do you measure the ROI?
You could measure reduction in response time. That’s easy to measure.
But what about customer satisfaction improvement? Did satisfaction increase because of the chatbot, or because you hired better support staff? Hard to separate.
What about the fact that the chatbot is learning from customer interactions, making future support better? That’s value, but it’s happening slowly. Over months. Over years maybe.
What about the fact that your support team is less burned out because the chatbot handles routine tickets? That improves retention. That improves morale. That probably improves the quality of support overall. But how do you measure it?
This is why only 31 percent of leaders think they’ll know within six months. Because the honest answer is: “It’s complicated.”
But companies need answers. CFOs need to justify budgets. Boards need to see ROI. So what actually happens?
Companies make up metrics. They call things “transformation” without measuring them. They announce AI projects as wins even though they’re still in pilot mode. They measure proxy metrics (like “we deployed 5 AI use cases”) instead of actual impact metrics.
This is dangerous because it means the good AI investments are being funded the same way as the bad ones. The winners look the same as the losers on a spreadsheet.
I think 2026 is when the reckoning happens. When companies stop being able to hide behind “we’re still measuring” and actually have to prove value or kill the projects.
The companies that figure out ROI measurement first will have a huge advantage. They’ll know which AI projects to double down on. Which ones to kill. Which ones to scale.
Everyone else will be throwing money at AI and hoping it works out.