The debate over self-management has a bad habit of turning managerial hierarchy into a villain. Bosses are cast as relics, layers as waste, the org chart as a monument to control. The research pack that a serious reader has to reckon with tells a duller and more useful story. Structure is contingent. There is no arrangement that wins everywhere, and the interesting question is not whether to have hierarchy but when.
The evidence refuses to crown a winner
Bernstein, Bunch, Canner and Lee (2016) open their assessment of holacracy by noting how polarized the field is: most observers take extreme positions, celebrating bossless workplaces or dismissing them as naive experiments. Their contribution is to refuse both. Foss and Klein (2022) reach a similarly deflationary conclusion from the data, finding that near-bossless firms are few and do not demonstrably outperform conventionally organized ones. If radical flatness were a general improvement, the performance gap would show up. It does not.
Lee and Edmondson (2017) supply the mechanism behind that null result. Managerial hierarchy works well for the reliable execution of known tasks in stable environments. Flat, self-managing designs, by contrast, earn their keep on complex, non-routine problems that cross functional lines. Neither claim cancels the other. They describe different jobs. Treating hierarchy as a moral failing is like treating a hammer as inferior to a wrench.
Managers do real work, not ceremonial work
The caricature of the manager is a person who exists to be reported to. The stronger view, drawn from strategy+business (2022) on the argument that managers matter, is that managers perform concrete functions: they design, organize, encourage, and enforce. Strip those out and the work does not vanish. Someone has to do it, formally or informally.
The same account draws a distinction worth carrying into any reorganization. There is authority of one kind, the power to hire, fire, and instruct, and authority of another, the power to lead, structure a problem, and build consensus (strategy+business, 2022). A firm can shed a great deal of the first while still needing the second. Flattening the chart often means transferring, not deleting, these functions, which is exactly why Foss and Klein (2022) keep finding informal bosses inside supposedly bossless firms.
The design, organize, encourage, and enforce functions that strategy+business (2022) attributes to managers are not optional overhead in a competitive market. Someone has to make sure the parts of the work connect, that commitments are kept, and that a decision once made is actually carried out. In a hierarchy these jobs sit in named roles. In a flat firm they are still done, but often by nobody in particular, and enforcement especially decays into whoever happens to be willing to nag. The question is never whether these functions exist. It is whether the organization has decided, on purpose, who performs them.
Match the structure to the conditions
The useful frame is contingency. Where work is stable and known, hierarchy is not just acceptable, it is efficient, and Semco itself kept a real organization even as it redistributed power (Semler, 1989). Where work is dynamic, uncertain, and cross-functional, tighter distribution of authority pays off, which is why W. L. Gore's lattice has sustained innovation for decades through direct person-to-person communication rather than a chain of command (Grønning, 2016). Most firms contain both kinds of work at once, which is why doctrinaire answers fail in practice.
The reversal at Zappos makes the contingency concrete. Groth (2020) documents a company that pushed all the way into holacracy, found it wanting, brought managers back, and settled on an internal market of teams with their own profit and loss. That is not a story of failure so much as one of calibration: the firm discovered which parts of its work needed named authority and restored it there.
The deeper reason no structure wins everywhere is that the trade-offs do not cancel. Foss and Klein (2022) find flat firms are few and do not demonstrably outperform, but they are careful not to claim hierarchy is superior in general either; their conclusion is contingent. Lee and Edmondson (2017), likewise, present less-hierarchical organizing not as a triumph but as a design with its own limits, one that demands more discipline, not less. When two independent strands of the research both refuse to name a universal winner, the sensible inference is that the winner is defined by the work, not by the ideology of the people arguing about it.
The practical implication for a leader is to stop asking whether the company should be flat or hierarchical and start sorting the work. Which decisions are routine enough to route up a clear line, and which are novel enough to need a team that can act without permission. Assign structure to the task, not to an ideology. A hierarchy imposed on genuinely uncertain work will smother it, and a flat structure imposed on stable, high-volume execution will make it slow and unaccountable. Most organizations already contain both kinds of work, which means the honest answer is almost never a single structure applied uniformly, but a mix calibrated function by function, and revisited as conditions change. The tool is not the enemy. Using the wrong one is.
Sources
- Bernstein, E., Bunch, J., Canner, N. & Lee, M. (2016). Beyond the Holacracy Hype. Frames the self-management debate as polarized and argues for a balanced assessment rather than celebration or dismissal. https://hbr.org/2016/07/beyond-the-holacracy-hype
- Foss, N. & Klein, P. (2022). Who's the boss? The persistence of entrepreneurial hierarchy in flat organizations. Finds near-bossless firms are few and do not demonstrably outperform, supporting a contingency rather than a winner. https://link.springer.com/article/10.1007/s41469-022-00132-8
- Lee, M. Y. & Edmondson, A. C. (2017). Self-managing organizations. Establishes that hierarchy suits stable known tasks while flatter designs suit non-routine problems, the mechanism behind the null result. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- strategy+business (2022). Can bossless management work? Distinguishes authority to hire, fire, and instruct from authority to lead and structure, and argues managers do concrete functional work. https://www.strategy-business.com/article/Can-bossless-management-work
- Groth, A. (2020). Zappos has quietly backed away from holacracy. Documents Zappos restoring managers and adopting an internal market, a case of calibrating structure to work. https://qz.com/work/1776841/zappos-has-quietly-backed-away-from-holacracy
- Semler, R. (1989). Managing Without Managers. Shows Semco kept a real organization while redistributing power, illustrating structure matched to conditions. https://hbr.org/1989/09/managing-without-managers
- Grønning, T. (2016). Working Without a Boss: Lattice Organization at W. L. Gore & Associates. Documents Gore's lattice sustaining innovation for dynamic work through direct communication. https://sk.sagepub.com/cases/download/person-to-person-communication-at-wl-gore-associates-inc