Executives are cutting layers, imposing return-to-office rules, and wondering if algorithms can do what middle managers used to. The risk is a false choice between hierarchy and chaos. Self-management offers a third path, but only when leaders replace supervisory control with explicit operating mechanisms and accept where the model does not fit.
What self-management is, and why it is hard
Michael Y. Lee and Amy C. Edmondson define self-managing organizations as those that radically, formally decentralize authority and eliminate the manager–subordinate relationship (Lee and Edmondson, 2017). Their synthesis argues that conventional hierarchy performs well for reliable execution in stable settings, yet tends to inhibit cross-functional problem solving and reinforces status differences that stifle people’s development. That combination explains the current tension. Firms face more problems that cut across boundaries, but many still rely on structures designed for repeatable tasks.
The same review also warns against naïve reform. Evidence on self-management is thinner than on traditional hierarchy, and most comes from cases rather than experiments (Lee and Edmondson, 2017). Leaders who crave a playbook will not find one. What they can borrow is a design principle: if you remove formal bosses, you must specify how decisions get made, how conflicts get resolved, and who maintains standards.
The operating system that replaces managers
Buurtzorg, the Dutch home-care provider, shows that scale without middle managers is possible when the work allows it. As of January 2021, Buurtzorg employed roughly 14,700 people organized in more than 900 self-managing teams, each up to 12 nurses, with only two directors and about 50 people in the back office (Martela and Nandram, 2025). The model relies on a thin coaching layer, just 21 internal coaches for about 950 teams, and those coaches have no formal authority or responsibility for team productivity. That choice is not cosmetic. It keeps decision rights with the teams and turns the coach into a support role rather than a hidden boss.
The economics are often misunderstood. A KPMG analysis around 2015 reported Buurtzorg’s overhead at 8 percent, compared with a 25 percent sector average in Dutch home care, but the company’s hourly cost was higher, €54.47 versus €48.74 (Martela and Nandram, 2025). In other words, the model did not simply “strip cost.” It shifted spending toward the front line and required a context where teams could operate with high autonomy.
That context matters. The Buurtzorg case identifies boundary conditions under which self-management travels better: lower interdependence across teams, highly motivated and competent front-line professionals, and services that can be effectively handled by small units with discretion (Martela and Nandram, 2025). These are not minor footnotes. They are placement rules. Try to lift the design into a setting with tight cross-team dependencies and you reintroduce coordination roles by stealth, or you drown teams in negotiation.
When process crowds out purpose
The Zappos experiment became a proxy war for beliefs about self-management. In 2016 it was the largest company to implement Holacracy fully, and the surrounding debate split between celebration and dismissal (Bernstein, Bunch, Canner and Lee, 2016). That polarization obscured a practical issue that any firm will face. If you swap line managers for intricate role governance and ritualized meetings, you risk building a process factory that chases itself.
The firm’s own arc is instructive. In March 2015, CEO Tony Hsieh issued an ultimatum to fully adopt Holacracy or take a buyout. Eighteen percent of employees left (Groth, 2020). By March 2017, Zappos had replaced Holacracy with an internal market in which teams operated as small businesses with their own P&Ls and transacted with each other at market-like prices. Over time it also brought managers back and shed the most bureaucratic meetings, keeping little more than a circular chart as an artifact (Groth, 2020). The move acknowledged two realities. First, rules-heavy self-organization can become inward-looking. Second, when coordination demands grow, you either add markets, add managers, or accept drift.
None of this makes Zappos a verdict on self-management. It is a reminder to pick the coordination mechanism on purpose. If you remove bosses and do not choose, the work will choose for you, and you may not like the result.
What it does to people
A common worry is that self-management simply shifts pressure from managers to teams and burns people out. Recent evidence complicates that view. Using two samples in Finland, one of 425 employees across multiple organizations and another nationally representative sample of 2,000, researchers validated separate scales for organizational self-management and individual self-management, then examined their links to well-being. Both forms of self-management were positively associated with work engagement and negatively associated with burnout, with relationships partially mediated by job demands (Morikawa, Martela and Hakanen, 2025). The study is correlational, not causal, and context likely matters a great deal. Even so, it challenges the claim that fewer bosses inevitably produce more stress.
The practical read
Start with the job of coordination, not the symbol of the org chart. If the work is modular, professional, and locally knowable, small autonomous teams with access to a coach can carry far more of the load than most firms assume, as Buurtzorg demonstrates at scale within explicit limits (Martela and Nandram, 2025). If the work is tightly coupled across functions, do not romanticize bossless designs. Either build market-like mechanisms, as Zappos eventually did, or design a minimal layer of integrators and be honest that these are managers by another name (Groth, 2020).
Then set the human guardrails. Define decision rights and standards clearly, publish conflict-resolution paths, and invest in coaching capacity that supports teams without commandeering them. Finally, measure outcomes beyond cost-to-manage. Track engagement and burnout alongside performance, and read the signals with care since the strongest recent evidence on well-being is associational rather than causal (Morikawa, Martela and Hakanen, 2025).
The choice is not hierarchy or anarchy. It is which coordination system fits your work today, and which one will still fit when the work changes tomorrow. Leaders who remove bosses and fail to install a working replacement are not pioneers. They are creating a vacuum. Leaders who choose their replacement system on purpose give their people room to solve real problems, and a structure that will not collapse the first time two teams need the same thing at once.
Sources
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines SMOs, explains when hierarchy helps or hinders, and flags the evidence base as largely case-driven. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Bernstein, Bunch, Canner and Lee (2016). Beyond the Holacracy Hype. Describes the polarized debate and Zappos’s then-status as the largest Holacracy implementation, framing the practical trade-offs. https://hbr.org/2016/07/beyond-the-holacracy-hype
- Martela and Nandram (2025). Buurtzorg: scaling up an organization with hundreds of self-managing teams but no middle managers. Documents Buurtzorg’s scale, coaching model without formal authority, boundary conditions, and reports dated overhead and higher hourly costs. https://link.springer.com/article/10.1007/s41469-024-00184-y
- Groth (2020). Zappos has quietly backed away from holacracy. Reports the 2015 ultimatum and 18% buyout, the 2017 shift to an internal market with team P&Ls, and the reintroduction of managers. https://qz.com/work/1776841/zappos-has-quietly-backed-away-from-holacracy
- Morikawa, Martela and Hakanen (2025). Self-management and employee well-being: validated scales, engagement and burnout. Finds that organizational and individual self-management correlate with higher engagement and lower burnout, with partial mediation by job demands. https://journals.sagepub.com/doi/10.1177/23409444241307734