Return-to-office fights, budget cuts, and AI pilots have the same collateral damage: they pressurize the layer in the middle. Managers report lower morale just as boards question the value of extra layers. The headcount math is already changing. Employers in the United States posted 42 percent fewer middle-management openings at the end of 2024 than in spring 2022, according to Deloitte (2025). At the same time, manager engagement fell sharply worldwide, from 27 percent to 22 percent between 2024 and 2025, part of a broader slide in employee engagement to 20 percent with an estimated economic cost of roughly 10 trillion dollars, or 9 percent of global GDP (Gallup, 2025).
The squeeze on the layer in the middle
Two forces are converging. First, complexity without clear value. More than two thirds of leaders see their organizations as overly complex and inefficient, and only half say they are prepared to anticipate and react to external shocks, a sobering baseline for any layer whose job is supposed to translate strategy into action (McKinsey, 2023). Second, a structural reset in how work is coordinated. Since the pandemic, about 90 percent of organizations have adopted some form of hybrid model that shifts coordination from supervision in a place to routines across places (McKinsey, 2023). When the work travels, so does the burden of proof for what a manager adds.
The numbers on middle managers themselves are not encouraging. In Deloitte’s global survey, managers say they spend nearly 40 percent of their time on administrative work and firefighting and only 13 percent developing their people. More than a third feel unprepared for the people part of the job (Deloitte, 2025). It is no surprise, then, that their engagement has sagged more than the average worker’s in the past three years (Gallup, 2025). The spread between typical and leading practice is striking. In best-practice organizations, 79 percent of managers report being engaged, almost four times the global average of 22 percent, which suggests engagement is a design outcome, not a macro inevitability (Gallup, 2025).
AI raises the stakes
Flattening is not just an austerity reflex. It is also a technology bet. Gartner forecasts that by the end of 2026, one in five organizations will use AI to flatten their structures. In those firms, more than half of middle-management positions could be eliminated, a scenario cited by Fortune as a forecast rather than a present fact (Fortune, 2025). Deloitte also reports the same one-in-five forecast on AI-enabled flattening by 2026 (Deloitte, 2025). Whether or not the percentage lands exactly as predicted, the direction of travel is clear: if routine reporting, coordination, and status tracking are automated, the least value-adding parts of the role will not survive. The question becomes what is left and how to make it measurably valuable.
Proof that a different shape can scale
The most credible counterexample to the idea that middle management is indispensable is not a Silicon Valley startup. It is a national home-care provider. Buurtzorg operates at significant scale without middle managers: more than 900 self-managing teams of up to 12 nurses and around 14,700 employees as of January 2021, supported by only two directors, about 50 back-office staff, and a network of 21 internal coaches who hold no formal authority and are not accountable for team productivity (Martela and Nandram, 2025). Historically, a KPMG comparison from around 2015 suggested Buurtzorg’s overhead was about 8 percent versus a sector average near 25 percent, a gap worth noting but not assuming unchanged today (Martela and Nandram, 2025). The model is not simply cheaper on every line, either. Buurtzorg’s hourly cost was higher than peers in the reported period, which undercuts the notion that flatness is just a wage squeeze (Martela and Nandram, 2025).
The case carries caveats that matter outside Dutch home care. The authors identify boundary conditions: the design works best when interdependence across teams is low, frontline professionals are both competent and motivated, and services can be modularized and governed by clear norms rather than constant escalation (Martela and Nandram, 2025). That is a far cry from a universal template, but it is a rigorous example of large-scale coordination without a conventional middle layer.
A decade ago, the management world took sides on this question. In their balanced reading of self-management experiments, including Zappos, Morning Star, Valve, and Buurtzorg, Bernstein, Bunch, Canner, and Lee argued against the polarized narrative that either celebrated “bossless” utopia or dismissed it as naive. They noted that in 2016 Zappos was the largest firm to have adopted holacracy in full, and urged leaders to parse where such designs actually fit rather than treat them as doctrine (Bernstein, Bunch, Canner, and Lee, 2016). That advice holds. The goal is not to copy a brand of self-organization. It is to redesign the middle of the organization to create value that software and process cannot.
A practical redesign, now
If the middle manager is melting, leaders have a choice about what refreezes in its place. Four moves follow directly from the evidence.
-
Strip the administrative ballast. If managers spend nearly 40 percent of their time on admin and firefighting, the fastest productivity gain is to remove it with shared services and AI-backed workflows, then track the time returned to coaching and problem solving (Deloitte, 2025). Use AI to collapse status reporting and coordination rather than to add dashboards that multiply work. The Gartner forecast on AI-enabled flattening is a warning shot that others will.
-
Redefine the job around development. Only 13 percent of manager time goes to developing people today (Deloitte, 2025). Make that the core, and measure it. The spread in manager engagement between typical and best-practice firms suggests design can change the experience and outcomes of the role (Gallup, 2025).
-
Replace layers with coach networks where the work allows. Buurtzorg shows that a small cadre of non-directive coaches, without formal authority, can support hundreds of teams when cross-team interdependence is limited and norms are strong (Martela and Nandram, 2025). Pilot this model in domains that fit those conditions instead of starting in the knottiest, most interdependent parts of the business.
-
Simplify spans and interfaces before you cut. Two thirds of leaders already judge their organizations overly complex and inefficient (McKinsey, 2023). Do not eliminate roles while leaving the tangle intact. First remove redundant handoffs and policies, then decide which managerial activities remain essential.
This is not a call for a mass bonfire of job titles. It is a call to treat the middle as design, not destiny. The organizations that will thrive in the next cycle are those that make the role smaller in headcount, bigger in craft, and clearer in purpose. The evidence already points to where the value is created: in teams that can run, leaders who coach rather than administer, and systems that handle the rest.
Sources
- Deloitte (2025). 2025 Global Human Capital Trends. Shows managers spend nearly 40% on admin and firefighting, only 13% on development, 36% feel unprepared, and that U.S. middle-management postings fell 42% since 2022, with a Gartner forecast that 20% of organizations will use AI to flatten by 2026. https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2025.html
- Gallup (2025). State of the Global Workplace 2025. Reports global engagement at 20% costing about $10 trillion (9% of GDP), manager engagement falling from 27% to 22% year over year, and 79% manager engagement in best-practice firms, underscoring design effects. https://www.gallup.com/workplace/659610/state-global-workplace.aspx
- Martela and Nandram (2025). Buurtzorg: scaling up an organization with hundreds of self-managing teams but no middle managers, Journal of Organization Design. Documents Buurtzorg’s scale without middle managers, its coach model and historical overhead comparison, higher hourly costs, and boundary conditions for when the model fits. https://link.springer.com/article/10.1007/s41469-024-00184-y
- Fortune (2025). Surviving the great flattening: the coming extinction of the middle manager. Cites Gartner’s forecast that by end-2026, 20% of organizations will use AI to flatten structures, eliminating more than half of middle-management roles in those firms. https://fortune.com/2025/09/19/surviving-great-flattening-coming-extinction-of-middle-manager-layoffs/
- McKinsey (2023). The State of Organizations 2023. Finds only half of leaders feel prepared for shocks, two thirds see their organizations as overly complex and inefficient, and about 90% adopted some form of hybrid work since the pandemic. https://www.mckinsey.com/capabilities/people-and-organizational-performance/our-insights/the-state-of-organizations-2023
- Bernstein, Bunch, Canner, and Lee (2016). Beyond the Holacracy Hype, Harvard Business Review. Argues for a balanced reading of self-management experiments and notes Zappos as the largest full holacracy adopter at the time, cautioning against polarized takes. https://hbr.org/2016/07/beyond-the-holacracy-hype