Two in five professionals now say their organization has cut back on the number of managers. Of the managers still in post, 55% say they are exhausted. And 39% of workers say the lack of managers leaves them feeling directionless (Korn Ferry, 2026).
Those three numbers come from the same survey of more than 16,000 professionals across 11 markets, released on 8 September 2026. Read together, they describe something more specific than burnout. They describe a transfer that never got written down.
Management layer cuts are modelled as a payroll line. The savings are real and they are easy to compute. What is harder to see on the model is that the removed layer was holding a portfolio of coordination, sequencing and priority-setting work, and that this work does not leave the building when the role does. It redistributes itself โ upward, sideways, and into the gaps between people who were never told they now own it.
What Korn Ferry Actually Measured
Precision first, because the press framing and the underlying instrument are not quite the same thing.
The headline reads "42% of organizations have cut management roles over the past year." The survey item behind it reads "My organization has cut back on the number of managers," agreed to by 42% of respondents (Korn Ferry Workforce 2026). That is a perception held by two in five professionals, not an audited count of organizations. It is still the most useful available signal โ employees notice when the layer above them thins โ but it is not a headcount statistic, and it does not contradict research showing median spans of control moving slowly.
The exhaustion figure deserves the same care. The item is "I feel exhausted by the pace of change at my organization." Across the whole sample, 49% agree. Among managers, it is 55%. The gap is the finding: the people nominally running the change are more worn down by it than the people receiving it.
Two more items from the same instrument matter for what follows. 61% of employees say they are performing the responsibilities of more than one role. Only 58% say they are appropriately rewarded for the additional roles they have taken on โ which means roughly two in five do not. Employee motivation fell from 71% in 2024 to 61% in 2026, a ten-point drop in a single survey cycle.
The Coordination Work Did Not Disappear. It Moved Up.
The strongest evidence that flattening relocates work rather than eliminating it is not new, and it is not survey data.
Julie Wulf tracked managerial job descriptions, reporting lines and pay structures across more than 300 established US firms over 14 years. Firms did delayer: the number of positions between division heads and the CEO fell about 25%, from 1.6 to 1.2, and the number of people reporting directly to the CEO rose from 4.5 in 1986 to almost 7 by 1999. The promise attached to that change was decentralization โ decisions pushed down, closer to the customer.
That is not what happened. Wulf found flattened firms exhibiting more control and decision-making at the top, with functional managers taking on more oversight and lower-level division managers playing a reduced role. Her summary is blunt: "Flattening can lead to exactly the opposite effects from what it promises to do" (Wulf, 2012).
A time-use study of 65 CEOs in the same research program showed the mechanism. Executives with broader spans did not work longer hours. They spent more time in meetings with multiple internal participants and less time alone. The coordination load did not vanish and it did not get absorbed for free โ it displaced something else on the calendar of the person who inherited it.
There is a parallel line of evidence on what that inherited layer was worth. Using production data from technology-based services workers, Lazear, Shaw and Stanton estimated supervisor effects and found them large: replacing a boss in the bottom 10% of quality with one in the top 10% raises a team's total output by more than adding a tenth worker to a nine-person team, and the average boss is about 1.75 times as productive as the average worker (Lazear, Shaw & Stanton, 2015). Workers assigned to better bosses were also less likely to leave.
That is the unpriced side of the trade. A manager removed from the chart is a salary saved and a productivity input withdrawn โ and the second number almost never appears next to the first.
Management Layer Cuts Hit Mid-Market Spans Hardest
If you run operations in a 50โ500 person company, the span-of-control data says the squeeze landed hardest on you.
Gusto's payroll and people-management data on small and mid-sized businesses shows the number of individual contributors per people manager roughly doubling โ about 3 direct reports in 2019, about 6 by 2024. The increase was not evenly distributed. Companies with 100โ499 employees saw the number of ICs per manager rise 44% between 2022 and 2024; companies with 2โ9 employees saw 24% (Gusto, 2025).
The flow data in the same analysis is the part worth sitting with. Since 2022, SMB hiring of people managers has declined by nearly half, while layoffs of people managers have surged threefold. Gusto also reports that sectors with proportionally more people managers tend to show higher labor productivity per hour โ a correlation, not a causal estimate, but one pointing the same direction as the supervisor-effects literature.
Mid-market companies did not run a formal delayering program. Most of them simply did not backfill. The structural outcome is identical and the decision was never documented, which is precisely why no one reassigned the work.
Then AI Landed on Top of the Gap
The sequencing here is what makes this a Q4 problem rather than a historical one.
52% of workers in the Korn Ferry sample say using AI has increased their workloads. 51% of individual contributors report improved efficiencies from AI โ against 79% of CEOs (Korn Ferry, 2026). The AI business case in most organizations is being validated 28 points away from the work it describes.
Add the structural context from the same survey: 62% report significantly heavier workloads over the past two years, 45% say they are too busy to deliver meaningful results, 44% say they are stretched beyond their capabilities, and 64% say their organization focuses more on driving efficiencies than on exploring new ways to add value. 52% say their performance is held back by existing processes and structures.
Deploying an agent into that configuration does not relieve it. Tools require specification, review and escalation paths โ all coordination work, all of it the kind of work the removed layer used to do. Automation introduced into an organization with no named owner for coordination does not reduce the coordination deficit. It adds a contributor to it.
The Honest Counter
Three limits, stated plainly.
This is self-report, not performance data. Korn Ferry measured perceptions โ exhaustion, directionlessness, motivation. No productivity outcome was observed. The 10-point motivation decline is a real movement in a consistent annual instrument, and it is not a measured output loss.
The flattening evidence is from a different population and era. Wulf's panel covers large US firms from 1986 to 1999 and concerns the top of the hierarchy, from division managers up to the CEO. A 200-person services firm cutting a team-lead layer in 2026 is not the same object. The direction of the finding travels; the magnitudes do not.
Not every cut layer was doing load-bearing work. Some management roles exist to review work that needed no review. Wulf notes the theoretical case that broader spans stop managers from micromanaging, and the 16% who say their organization is ineffective at turning ideas into outcomes is low enough to suggest the problem is rarely a shortage of supervision per se. The argument is not that flattening is wrong. It is that flattening without an explicit reassignment of the coordination portfolio is an unpriced liability.
What to Do Before the Next Layer Goes
Four moves, and none of them requires reversing a cut.
- Inventory the orphaned decisions, not the orphaned tasks. For each management role removed in the last 18 months, list what that person decided or sequenced weekly: prioritisation calls, cross-team handoffs, exception approvals, quality gates. Tasks get noticed when they drop. Decisions just quietly stop being made.
- Name one accountable human per orphaned decision. Not a committee, not "the team." The Korn Ferry item that should alarm you most is the 39% feeling directionless โ that is the symptom of a decision with no owner, and it is fixed by writing a name next to it.
- Price the stretch or end it. Two in five say they are not appropriately rewarded for the additional roles they have taken on. A permanent stretch that was framed as temporary is a retention liability with a known mechanism.
- Resolve the coordination gap before you add a machine to the seat. If you are considering AI or agents in a thinned structure, the prerequisite is knowing which seat holds which decision and who approves what. Scovai's AUGMENT org-chart model scores each seat for augmentability and outputs the oversight chain alongside it โ a reasonable reference for the shape this mapping needs, whoever builds it. An agent dropped into an unowned coordination gap inherits the ambiguity, not the authority.
The savings from a management layer cut show up in the quarter you make it. The coordination debt shows up two quarters later, distributed across people who never saw the decision and cannot name what changed.
Before you approve the next structural cut, ask one question of the plan: which specific decisions does this person make, and whose name goes next to each one on Monday? If the plan cannot answer, you are not removing a cost. You are deferring one.
