A tech CEO’s belief that AI-enabled “self-managing teams” will make middle managers unnecessary collides with survey and research data: managers drive at least 70% of employee engagement variance, and past experiments show that removing the role redistributes
A head of HR at a tech company asked a question that sounds efficient—and terrifyingly final: “Are you hearing of other organizations eliminating the manager role?”
The CEO’s view was blunt. AI, the executive argued, will make middle management redundant. The future will be “self-managing teams.” In the room, it landed with the awkward weight of a decision that can’t be undone once people leave.
This isn’t a niche worry. Some 41% of employees say their companies trimmed management layers last year, according to Korn Ferry’s survey of 15,000 professionals worldwide. Middle managers accounted for more than 31% of all layoffs in 2023. And the momentum appears set to continue: Gartner predicts that through 2026. 20% of organizations will use AI to flatten their structure. eliminating over half of their middle management positions.
The appeal is real. Management layers are expensive, and bad managers do exist. If AI can take meeting notes, draft goals, schedule one-on-ones, and flag underperformers in a dashboard, the logic goes, why keep a human in the middle?
But the data the CEO ignores points in the other direction—straight through the fantasy that the work of management is mostly paperwork.
Gallup’s research, for example, puts a number on what executives often feel but rarely quantify: managers account for at least 70% of the variance in employee engagement scores across business units.
Seventy percent. Not perks. Not even the words executives deliver in all-hands.
The real lever is the person your team member reports to—because that manager shapes the day-to-day conditions that determine how much effort people choose to bring. Engagement, Gallup says, connects to customer ratings, profitability, productivity, quality, turnover, absenteeism, theft, and safety incidents. Those are not abstract outcomes. They’re the metrics boards and CEOs actually chase.
So when a CEO asks, “Do we even need managers?,” the question beneath it is harsher: can the organization cut 70% of the variance in results—and trust that everything still moves the right way?
That’s a big bet.
Companies have tried to make the bet before. using different names: flat structures. Holacracy. and “self-management.” Zappos is the most famous example. More than a decade ago. it eliminated the manager role and gave employees the option to stay or take a buyout. Within a year, 210 staff—14% of their workforce—took the buyout. But only 20 of those were managers. That meant 190 individual contributors voluntarily decided to leave.
In the years after, Zappos ran into “big challenges” and brought managers back.
Another early adopter, Medium, later abandoned its own self-management approach after the system began taxing its team’s effectiveness and sense of connection.
The pattern that follows these experiments is almost always the same. When companies remove the manager, they don’t remove management. They redistribute it. And in the new arrangement. senior executives absorb more direct reports and operational tasks. leaving humans who were once protected by a clear management structure questioning whether they can actually do the job they’ve been moved into.
The work doesn’t vanish. Accountability and coaching do.
That delay matters. High-potential employees aren’t learning how to make tough calls anymore—and the leadership crisis created by that gap won’t show up until about two years after the cuts.
If the pitch shifts from removing managers to replacing them with AI, the stakes get sharper. In the AI version of the fantasy, the “manager” becomes a dashboard. Performance is scored. Any feedback arrives as a summary.
The difference shows up in exit interviews—because employees don’t describe “systems” when they leave. They describe moments.
They say: “Nobody told me what was expected.”
They say: “I didn’t feel like I mattered.”
They say: “I had no idea if I was doing a good job or about to get fired.”
None of those get solved by an algorithm.
They get solved by a human who knows your name, your pet’s name, and what you’re trying to build in the work—and in your career.
It’s not a job an AI can do alone.
Org charts and spreadsheets don’t show what leads to engagement. They don’t capture the human friction inside the day.
The high performer who used to stay late and comes back from maternity leave—nursing in the parking lot between meetings—convinced she’s failing at her job and failing as a mom.
The team member who receives not-so-great biopsy results for his mother the previous Tuesday, not telling anyone, then appearing at 9 a.m. in a Zoom meeting with the camera off.
The quiet team member holding it together because her closest work friend just got laid off—balancing guilt for still having a job with the stress of being expected to do the job of two people.
This isn’t “we’re a family here.” An employment lawyer would discourage that phrase. The point is measurable: how a manager handles a 30-second conversation can affect whether someone sleeps tonight—and the work they do all year.
A manager who prevents the lawsuit, who gives clear feedback, who develops people—those aren’t theoretical wins. They show up as reduced risk, reduced regretted attrition, and a pipeline that doesn’t collapse.
A manager who can read the room and have the hard conversation can prevent the lawsuit. The employment-law lesson embedded in the caution is specific: “Well, nobody ever actually told her that her performance was a problem.”
A manager who gives clear feedback can prevent the regretted attrition. The real-talk version of why people stay is straightforward: “My boss is great—why would I ever leave?” The opposite scenario plays out just as clearly: the person whose manager never showed up for their one-on-one ended up spending that time updating their résumé instead.
The cost of replacing an employee runs anywhere from half to two times their salary, depending on the role. Multiply that by turnover driven by disengagement—the kind that turns into résumé-updating—and the “expensive middle layer” can pay for itself.
Managers also build what comes next. But the pipeline is already thinning. Deloitte research found that only 6% of Gen Zers say achieving a leadership position is their primary career goal. The logic here is painful: when young workers watch middle managers get eliminated for efficiency and conclude the path isn’t worth it. fewer people commit to leadership.
AI may be excellent at many tasks. It doesn’t develop a 24-year-old into a 34-year-old who can run a P&L.
For CEOs weighing whether they need managers, the question being asked is the wrong one. The more important question is whether the organization has the right managers—and whether it has trained them.
Gallup’s research adds another blunt fact: companies miss the mark on high managerial talent in 82% of their hiring decisions. The pattern behind that number is familiar—promote the best individual contributor and assume that talent will automatically transfer to leading a team. without providing training to do so. When that doesn’t work, frustration builds, and people leave.
The answer isn’t fewer managers. It’s better-supported ones: people trained to set expectations, give feedback, handle moments that matter, and drive results without driving their team out the door.
For the people already doing the job, the message is direct. For the CEOs asking whether the role should disappear, it is just as direct.
This isn’t theoretical research. It’s the real business outcomes tied to the human on the other side of a next one-on-one—whether they say it out loud or not.
middle management AI and leadership organizational flattening employee engagement Gallup research Korn Ferry survey Gartner prediction layoffs 2023 Zappos buyout Holacracy self-management