The AI productivity stories are everywhere. The financial results don’t always tell the same story.
McKinsey’s 2026 global survey explains the gap almost perfectly. Eighty percent of respondents said AI had improved their individual productivity. Only 37 percent attributed any positive EBITA impact to AI at the company level. Just 6 percent qualified as AI high performers, meaning they attributed at least 5 percent of EBIT to AI and described its impact as significant.
That does not mean the other 63 percent received nothing. Respondents also reported improvements in innovation, competitive differentiation, customer satisfaction, and employee satisfaction, as well as financial gains in specific business functions. This was also a survey, not a forensic audit of everybody’s income statement.
Still, the pattern is pretty hard to miss.
People got faster. For most organizations, the P&L did not keep pace.
If you are honest with yourself, I bet you know the feeling.
Personal Productivity Is Easy to Feel
AI has given me more professional energy than almost any technology before it. It can help me read, research, organize, compare, question, create, and occasionally find the weak link at a speed that would have seemed impossible a few years ago.
That improvement is easy to feel. A task that took two hours takes 20 minutes. A leader walks into a meeting better prepared. An employee gets through a first draft before lunch instead of staring at a blank document until Thursday.
All of that matters. But personal productivity and enterprise productivity are not the same thing.
An employee can finish a task faster and then wait three days for the same approval. A manager can create a better recommendation, send it into the same six-person committee, and watch it sit untouched for another week. The person improved. The process did not.
We have automated the sprint and still preserved the traffic jam.
Saved Time Is Capacity, Not Yet Value
Leverage is what management decides to do with that capacity. Will the employee complete more work? Will the customer receive an answer faster? Will quality improve? Will mistakes decline? Will the company absorb more volume before adding the next ten people? Will the saved time help launch a new product or solve a problem nobody previously had enough capacity to address?
Or will everyone simply feel a little less busy while the same meetings, reports, approvals, handoffs, and economics remain exactly where they were? Is quiet quitting really quitting if you enable your team to work 1/10th of the time they did before the introduction of AI? Sounds like a new strategy we should all be on the lookout for.
The P&L does not automatically recognize that an employee saved 20 minutes unless you are embedding the efficiency gain into the process and overall strategy. Somebody has to tell those 20 minutes where to go.
The Workflow Has to Change
The high performers in McKinsey’s survey were not merely prompting better. Nearly three-quarters reported fundamentally redesigning workflows around AI, compared with only one-quarter of the other respondents. They were also twice as likely to report clear commitment from senior leaders and defined processes for measuring the impact of AI initiatives.
That is the difference. They did not simply insert AI into the old work. They changed how the work moved.
If AI creates the first draft, maybe someone else should stop creating the same first draft manually. If AI validates the documentation, maybe every file no longer needs to travel through the same review process. If AI identifies the exceptions, perhaps trained people should spend their time on the exceptions instead of repeatedly approving the obvious.
If this sounds familiar, it is because it is the same challenge of creating true scale across workflows, validation, and end-to-end transactional automation.
You cannot add a new tool, preserve every piece of old work, and then act surprised when the economics fail to move.
Instead of transformation, all you’re getting is accumulation. With a lack of leadership and accountability in your teams, you are going to see a whole lot of red lights through the front windshield on your journey.
Measure the Business, Not the Magic Trick
The question is not whether AI can create something useful in 30 seconds. I’ll save you a token: It can and does.
The question is what changed after the answer appeared. Did cycle time improve? Did output increase and create tangible dollar value? Did the quality of your product’s features and benefits improve? Does the customer even care? Did the company handle more work and bill for it? Did cost actually decrease on the very bottom line? Did the organization learn something it can repeat across the company?
An amazing demonstration is not an operating result. A faster employee is not automatically a faster company. A thousand individually impressive AI moments can still produce one very ordinary enterprise outcome if the workflow around them never changes.
We all know that AI works. The evidence for that keeps getting stronger.
But AI does not carry productivity from an employee’s desk to the company’s financial statements. Leaders still have to redesign the work, decide what the saved capacity is for, and remove the friction surrounding the newly improved task.
People getting faster is promising.
The company getting better—measured by actual dollars—is the point.
—
Originally posted on Forbes.com



