Why Personal AI Gains Do Not Automatically Reach the Bottom Line

A machine operator who works visibly faster with an AI-assisted tool is a good sign, and the hope that it will show up in the annual accounts is a fair one. A joint study by MIT Sloan Management Review and Boston Consulting Group, published in 2022, puts a number on the link: organizations whose employees personally derive value from AI are 5.9 times as likely to report significant financial benefits from AI as organizations whose employees get no value from it. Mid-sized companies rarely introduce AI through a dedicated research unit and instead do it tool by tool in daily operations. For them that is a practical point of orientation.

Where the Number Comes From

Behind the study sits a global survey of 1,741 managers across more than 20 industries and 100 countries, complemented by 17 in-depth interviews. For a fair reading, it matters who answered: managers describing their own organization, rather than an auditor with access to the books. The study therefore measures what people inside companies perceive, and it shows what occurs together. Whether personal value causes the financial results, or whether both trace back to a third cause such as a management team that introduces new things carefully, remains open. The finding still carries far in practice: when companies with satisfied users report results that much more often, it is worth asking whether your own users gain anything from the tools.

Four Fields, One Map

The study sorts the answers onto a board with two axes, much like the supplier scorecards that are common in purchasing. One axis asks what an individual employee personally gains from AI, such as time saved or more confidence in their own task. The other asks what the organization gains. Four fields emerge, and the split rewards a second look: 37 percent of respondents report at least moderate value on both sides, while 30 percent see no meaningful value on either. Between these two extremes sit the mixed fields, and those are the more interesting ones for management. 26 percent feel a personal benefit that never reaches the company result. 7 percent describe the reverse, organizational value without a visible gain for employees.

A word on "at least moderate", because the reading depends on it: the study counts into this field everyone who rates the value as moderate, significant, or extensive. The 37 percent are therefore no circle of showcase operations; they reach down to companies where things run noticeably better in an unspectacular way. That is the friendlier message in the number: value you notice in daily work is enough to reach this field, and a lighthouse project is optional.

For a leadership team wondering why an AI rollout has not yet moved the numbers, the more likely answer sits in the two middle fields: the value exists, it simply arrives on one side and stalls on the other.

When the Gain Does Not Survive the Shift Handover

Every plant knows the trick one shift has worked out for itself and that never survives the handover. It saves the person who knows it 20 minutes a day, and it appears in no process description. The moment that person goes on holiday it is gone. This is exactly the pattern behind the 26 percent: 20 minutes saved in the head of one clerk are a real gain for her. They stay invisible to the company as long as the surrounding process remains unchanged.

Why does that happen? At this point the study names a cause that is easy to recognize at home: attention beats impact. The report quotes a former manager at McDonald's, Dave Galinsky. He cautions against focusing on the shiny things in AI and machine learning instead of on actual customer value and the customer's experience with the company. Machine learning here simply means software that derives its rules from examples rather than having them written out in advance. A tool that impresses in a demo can miss precisely the spot where money is earned: quote turnaround, spare-part availability, the response to a breakdown.

The way out of this field runs through the small, genuine benefit that becomes visible early and is written into the process. The trick of one shift becomes the rule of the house, and 20 minutes saved per person becomes a figure that appears in the costing.

The Flip Side: Value Without Participation

The rarer case covers 7 percent, a small group, yet it deserves more attention than its size suggests. Here the organization profits while employees come away empty-handed. The report describes a recurring pattern behind it: effort and reward sit in different places. Whoever maintains the data so that a system can make useful suggestions carries the work. Another department takes the gain, sometimes two levels up. The report puts this twice: as a warning against burdening workers to serve the machine, and as the requirement that the benefit for an individual must offset that individual's effort.

This constellation describes the projects that work on paper and find no advocate on the floor. It feeds a skepticism that resurfaces at the next AI initiative, that time with less willingness to engage. Balancing the effort where it actually falls therefore buys two things: participation today, and the readiness to come along again with the next tool.

What the Report Recommends to Leadership

Against the gap between personal and organizational value, the report offers four recommendations:

• Resist the lure of shiny AI projects that carry little real value.
• Look for benefits on both levels from the outset, for the individual employee and for the organization.
• Keep employees from being reduced to servants of the machine.
• Make sure the benefit for an individual offsets that individual's effort.

Elsewhere the report describes what managers can work on day to day: building trust in the tool, leaving employees room to decide how they use it, and being open about when and how it is used. The last point sounds like a formality and is the opposite of one. Someone who knows what a system makes suggestions for, and where it stays out of the way, will work with it instead of around it. That is the difference between a tool that lands in the business and one that stays in a folder.

Value Counts on Both Sides

The study's map describes four states, and it says nothing about which one is yours. That can be settled in an hour, in scheduling, in purchasing, or in customer service. Three questions are enough: who does the work with the new tool? Who takes the gain from it? And where is that gain written down, so that it outlasts the shift handover?

Our position matches the study's finding on one point: value holds only when it is thought through from both sides. The employee gains time or confidence in her own task, the company gains the shorter quote turnaround or the more precise maintenance schedule that eventually reaches the customer. With one of the two sides missing, the value stays parked in the 26 percent field or the 7 percent one, and there it pays nothing in. Both sides together mark the field where 37 percent of respondents already stand, and they are the reason an AI rollout adds up at all.

The basis is the study "Achieving Individual and Organizational Value With AI" by MIT Sloan Management Review and Boston Consulting Group, published in 2022, with a global survey of 1,741 managers across more than 20 industries and 100 countries plus 17 interviews. The 5.9x factor, the 37, 30, 26, and 7 percent figures, the four recommendations, and the paraphrased statement by Dave Galinsky all come from the report. "At least moderate value" is the report's own counting rule, which groups moderate, significant, and extensive value into one field. The figures are self-reported by respondents and describe a correlation rather than proof of cause and effect. The survey dates from 2022, before today's language models became widely available, and the numbers should be read accordingly. The focus on mid-sized companies and the shift-handover image are our own framing.