The debate about middle management usually happens in the language of culture and morale. Layoffs season turns it into something colder and more quantitative. When a company decides to remove a tier of managers to cut cost, it is implicitly making a claim about how much that tier was worth. The self-management cases let us put real figures on both sides of that ledger, and the numbers are more interesting than the rhetoric.
The overhead is countable
The cleanest figure comes from Dutch home care. Martela and Nandram (2025) report that Buurtzorg ran with an overhead of roughly 8 percent against a sector average near 25 percent. That gap is the price of coordination, and it is not a rounding error. It is worth being careful, though: that comparison traces to a KPMG assessment from about a decade ago, so it is a past snapshot rather than a standing law, and it should not be quoted in the present tense. The structural facts behind it are sturdier. Martela and Nandram (2025) describe an organization of about 14,700 people run by two directors, some fifty back-office staff, and 21 coaches serving roughly 950 teams. Whatever the exact overhead ratio today, an administrative footprint that thin is a design choice with obvious cost implications.
The honest complication is that low overhead is not the same as low cost. Martela and Nandram (2025) also note that Buurtzorg's cost per hour of care has run higher than a comparison benchmark, around 54 euros against roughly 49. The two facts coexist because the savings do not come from paying nurses less. They come from spending fewer hours per patient and producing better outcomes, so the lean administrative structure translates into value through effectiveness rather than through a cheaper unit price. Anyone building a business case for delayering should copy that nuance, because a board that expects self-management to lower wages is being sold the wrong story.
That distinction between overhead and unit cost is where most delayering programs go wrong before they start. A firm that removes a management tier and keeps everything else the same has cut overhead on paper while leaving the actual work of coordination unassigned. The Buurtzorg figures only make sense as a package: the thin administrative footprint and the effectiveness gains are two outputs of one design, not two levers a CFO can pull separately. Take the overhead number without the operating model that produced it and you have a spreadsheet fantasy, not a plan.
The other side of the ledger
Against overhead sits the growth argument, and here the evidence is softer and must be labeled as such. Hamel (2011) reports the Morning Star founder Chris Rufer's claim that the company's volumes, revenue, and profit grew at double-digit rates over twenty years against a sector averaging around one percent a year, with expansion financed almost entirely from internal funds, all inside a firm with no managers, no titles, and no promotions. That is a striking figure, but it is a founder's self-report inside a case study, not an audited or independently verified result, and it would be dishonest to present it as proof that removing managers caused the growth.
What can be said more firmly is structural. Hamel (2011) argues that management is itself a cost, and not only in salaries. Each layer of approval taxes speed, filters information, and dilutes the initiative of the people below it. Kanter and Dai (2018) describe Haier acting on the same logic from the top down, restructuring into microenterprises whose platform owners were explicitly not supervisors, so that the discipline came from the customer rather than from an expensive internal chain of sign-offs. The cost of the middle layer, in this reading, is partly the payroll and partly the drag.
What a CFO should actually take from this
The temptation during a cost-cutting drive is to remove managers and pocket the salary savings. The cases suggest that is the least interesting part of the number. Martela and Nandram (2025) show that the prize is a fundamentally leaner coordination structure, not a one-time headcount cut, and that it only pays off when the work suits it: low interdependence between units, a capable and motivated frontline, and services that reward local judgment. Strip out managers without meeting those conditions and you keep the coordination problem while removing the people who were solving it.
So the financial case for self-management is real but conditional. Overhead in the single digits is achievable, as Buurtzorg shows, and the drag of excess approval layers is a genuine cost, as Hamel (2011) and Kanter and Dai (2018) both argue. But the savings are earned through effectiveness and speed, not through cheaper labor, and they depend on the same boundary conditions that determine whether the model works at all. A CFO who understands that will make a stronger, more durable case than one who simply totals up the salaries of the layer about to be cut.
Sources
- Martela, F. and Nandram, S. (2025). Buurtzorg: scaling up an organization with hundreds of self-managing teams but no middle managers. Source for the ~8% versus ~25% overhead figure (a decade-old KPMG snapshot), the thin administrative footprint, and the higher cost-per-hour caveat. https://link.springer.com/article/10.1007/s41469-024-00184-y
- Hamel, G. (2011). First, Let's Fire All the Managers. Provides Rufer's self-reported growth claim and the argument that management is a cost in salaries and in drag. https://hbr.org/2011/12/first-lets-fire-all-the-managers
- Kanter, R. M. and Dai, N. H. (2018). Haier: Incubating Entrepreneurs in a Chinese Giant. Shows a top-down effort to replace supervisory sign-offs with market discipline. https://www.hbs.edu/faculty/Pages/item.aspx?num=54098