Two organisations removed their managers. One is cited as proof that it works and one as proof that it does not, and the pair is more useful than either on its own.
The Dutch home-care provider Buurtzorg is the success story. Zappos is the other one.
What Zappos actually did
The online shoe retailer, owned by Amazon since 2009, began adopting Holacracy in 2013. Holacracy is a specific, written constitution for running an organisation without managers: work is organised into circles with defined roles rather than job titles, decisions follow a formal governance process, and the authority sits with the role rather than with the person.
By the beginning of 2015 about 85 percent of the company had made the transition. In April of that year the chief executive, Tony Hsieh, sent a memo offering every employee a severance package if they preferred not to continue. To take it, employees had to be in good standing, and they were asked to confirm by email either that they had read the management book Reinventing Organizations and disagreed with its manager-free argument, or that they were not reading it.
210 of 1,503 employees took the package before the 30 April deadline. That is close to 14 percent of the company, and Zappos confirmed the figure to the Las Vegas Sun at the time. Later rounds took the total higher, and contemporaneous reporting put it at around 18 percent by early 2016.
What that number does and does not tell you
It is the number everybody quotes, and on its own it establishes less than it appears to.
There is no comparison group. Zappos had double-digit turnover before Holacracy, as most retail and contact-centre operations do. Without knowing what attrition would have been in the same year without the change, 14 percent is a figure without a baseline.
The offer was engineered to produce departures. A severance package conditional on stating that you disagree with the direction is a self-selection mechanism, and it worked as designed. Reading the resulting number as a referendum on Holacracy misreads what was being measured: it counted people willing to take money to leave, at a company already known for paying people to leave during onboarding.
It says nothing about performance. No independent evaluation of Zappos under Holacracy was ever published. There are no comparable figures on productivity, service quality or cost, because nobody was commissioned to produce them.
That last point is the interesting one, and it is the difference between this case and the Dutch one. Buurtzorg's claims were checked because the Dutch Ministry of Health commissioned KPMG to check them, after competitors complained. Zappos was a private subsidiary running an experiment on itself, and the entire public record is journalism and the company's own statements.
Two famous cases about the same idea, and only one of them has an evaluation. When a management practice is described as evidence-based, that asymmetry is usually what is underneath it.
The retreat, and who wrote the warning
Zappos did not announce an end to Holacracy. It moved away from it gradually: the rigid governance meetings went, elements of management came back in a different shape, and by 2020 reporting described a company that retained a circular structure and little of the formal constitution.
The most useful document to come out of the episode was published in the middle of it. In July 2016, Harvard Business Review ran Beyond the Holacracy Hype, subtitled "the overwrought claims and actual promise of the next generation of self-managed teams", by Ethan Bernstein, John Bunch, Niko Canner and Michael Lee.
Its argument, drawn from multi-year research at several organisations that had adopted these forms, is that self-management can make an organisation more adaptable but that most companies should not adopt it wholesale. The recommendation is piecemeal: self-management where the need for adaptability is high, conventional structures where reliability is what matters.
One of the four authors is John Bunch, who led the Holacracy transition at Zappos. A person running the most-watched implementation in the world co-authoring the case against implementing it wholesale is not a contradiction. It is the most credible thing in the entire record, and it is almost never mentioned when the case is told.
What separates the two cases
Set the two side by side and the difference is not national culture, or leadership conviction, or how committed the employees were.
| Buurtzorg | Zappos | |
|---|---|---|
| Self-management from | Founding, 2006 | Retrofitted from 2013 |
| Surrounding systems | Built for it: IT, billing, back office | Existing systems and an owner's reporting lines |
| The work | Nursing, high autonomy, low interdependence between teams | Retail and contact centre, high interdependence |
| Independent evaluation | Government-commissioned, KPMG 2015 | None |
| Where it stands | Continuing | Stepped back |
The two structural differences are the ones that carry weight. Buurtzorg was designed around self-management from the first day, so its IT, its billing and its back office exist to let twelve nurses run their own administration. Zappos layered a constitution onto a company that already had systems, reporting lines and an owner.
And the work is different in a way that matters. A nursing team's output is largely independent of the team next door. A retail operation's is not: fulfilment, merchandising and service are tightly coupled, and coupling is exactly what a coordination structure exists to handle. Removing the structure without replacing the coordination leaves the coordination to be improvised.
That is the same conclusion the scoping review of 25 Buurtzorg copies reached from the other direction: the obstacles were the surrounding organisation and the policy environment, not the willingness of people to work autonomously.
What to take from it
The failure was not of self-management. It was of retrofitting one specific, highly formalised version of it onto an existing operation with tightly coupled work, without changing what sat around it.
The departure figure is not the finding. It is the most quoted number in the case and the weakest one in it.
Nobody measured the thing that matters. Five years of the most closely watched organisational experiment of its decade produced no published data on whether the company got better at anything. If that is true of the famous case, assume it is true of the one in the deck you are being shown.
Neither of these cases supports a general claim about hierarchy. What the pair supports is narrower and more useful: how work is coupled, and what sits around the teams, decides whether removing the managers is a structural change or just a removal. The general literature on this is thin in a way that is worth knowing about, and which change model has evidence behind it sets out how thin.
Common questions
Why did Zappos adopt Holacracy? Tony Hsieh wanted the company to behave more like a city than a corporation, on the argument that cities get more productive as they grow while companies get less so. The adoption began in 2013 and about 85 percent of the company had transitioned by the start of 2015.
How many employees left Zappos over Holacracy? 210 of 1,503, close to 14 percent, took a severance package before the deadline of 30 April 2015, a figure Zappos confirmed at the time. Later rounds raised the total, reported at around 18 percent by early 2016.
Does that prove Holacracy failed? No. There is no comparison group, Zappos had double-digit turnover before the change, and the offer was explicitly designed to let people who disagreed take money and leave. The figure counts self-selection, not outcomes.
Did Zappos abandon Holacracy? Not by announcement. It stepped back gradually, dropping the formal governance meetings and reintroducing elements of management, while keeping a circular structure. Reporting in 2020 described a company that had quietly moved on from the strict form.
Is there any evaluation of Zappos under Holacracy? No published independent evaluation exists. The public record is contemporaneous journalism and the company's own statements, which is the main difference between this case and Buurtzorg.
What does the research say about self-managing organisations? The most cited assessment, in Harvard Business Review in 2016, argues that these forms can make organisations more adaptable but that most companies should not adopt them wholesale, and recommends applying self-management where adaptability matters and conventional structures where reliability does.
Why did it work at Buurtzorg and not at Zappos? Two structural differences. Buurtzorg was designed around self-management from its founding, so the systems around the teams support it. And nursing teams work largely independently of each other, while retail and contact-centre work is tightly coupled, which is what a coordination structure is for.
Sources
- HR Dive (10 May 2015), 14% of Zappos employees cash out after radical self-management plan, reporting figures Zappos confirmed to the Las Vegas Sun
- Bernstein, Bunch, Canner & Lee (2016), Beyond the Holacracy Hype: The overwrought claims and actual promise of the next generation of self-managed teams, Harvard Business Review 94(7-8), 38-49
- Quartz (2020), Zappos has quietly backed away from holacracy
- Gray, Sarnak & Burgers (2015), Home Care by Self-Governing Nursing Teams: The Netherlands' Buurtzorg Model, The Commonwealth Fund
