The financial case for removing middle management is usually made with one striking comparison and a lot of hope. The comparison is worth taking seriously. The hope is where flattening programs go wrong, because the full accounting is more interesting, and more useful, than the pitch.
Start with the number that makes the case. Buurtzorg, the Dutch home-care company that runs more than 900 self-managing teams and roughly 14,700 staff with no middle managers, has operated at an overhead of around 8 percent, against a sector average closer to 25 percent in Dutch home care (Martela and Nandram, 2025). That is not a rounding difference. An organization spending 8 cents on overhead where competitors spend 25 has redirected a large fraction of its cost base from supervision to service. The two directors and roughly 50 back-office staff supporting nearly 15,000 people are the visible proof that a thin center can hold (Martela and Nandram, 2025).
The part the pitch skips
Here is where honesty separates a real business case from a slide. Lower overhead did not make Buurtzorg's care cheaper per hour. By the analysis the company is best known for, its cost per hour of care came in higher than a large comparator, roughly 54 euros against 49 (Martela and Nandram, 2025). The flat structure bought something real, but it was not a lower unit price. It was the ability to spend more of each hour on care and less on managing the people delivering it, which shows up as quality, continuity, and retention rather than as a cheaper invoice.
This distinction matters because most flattening decisions are justified as unit-cost savings. Remove the managers, the reasoning goes, and each unit of output gets cheaper. Buurtzorg's own numbers refuse that logic. Overhead fell dramatically and unit cost did not, which means the manager's salary was not simply dead weight sitting on top of the price. It was buying coordination that, when redistributed to well-paid frontline staff, costs real money in a different column. Flat moved the cost. It did not conjure it away.
What boards are actually buying
If the saving is not in unit cost, where is the return? It is in the things overhead-heavy organizations quietly lose. Gallup's 2025 survey put the global cost of disengagement near 10 trillion dollars and recorded manager engagement falling from 27 percent to 22 percent in a year (Gallup, 2025). A structure that keeps frontline staff engaged and close to their work is buying down that loss, even if it never shows on the overhead line. Deloitte's finding that managers spend nearly 40 percent of their time firefighting points at the same waste from the other side: much of the cost of a hierarchy is not the salaries but the friction the layers generate (Deloitte, 2025).
So the genuine business case for flat is a portfolio, not a single number. Lower overhead is the part that fits on a spreadsheet. Higher engagement, faster decisions, and better retention are the parts that fit on the next few years of results. A board that approves flattening expecting an immediate drop in unit cost will be disappointed, and will likely blame the model rather than the pitch. A board that approves it understanding the trade, more spent per hour on the actual work in exchange for far less spent on managing it, is buying the right thing.
Retention is the line where the portfolio pays off most concretely. Every departure carries a replacement cost in recruiting and lost productivity, and a structure that holds people reduces how often that bill arrives. Gallup's best-practice organizations reach 79 percent manager engagement against a 22 percent average, a gap that translates directly into who stays (Gallup, 2025). That is the return a flat model earns quietly, invoice by invoice that never gets sent, and it is exactly the return an overhead-only reading of the case fails to count.
The honest read
The temptation in the great flattening is to sell it as free money, because the overhead comparison is so dramatic and the layoff so immediate. Buurtzorg, the most studied example of the model working, quietly warns against that framing with its own accounts. The overhead saving is real and large. The unit-cost saving is not there, and the company never claimed it was.
There is one more line in the honest ledger. The Buurtzorg figures come with caveats worth stating plainly: the overhead comparison rests on an analysis from the mid-2010s, the headcount is from early 2021, and one of the researchers has documented ties to the company (Martela and Nandram, 2025). None of that overturns the case. It disciplines it. Flattening well is a bet that you can convert supervisory overhead into engagement and quality without inflating the cost of the work itself. Ricardo Semler made a version of that bet for decades and defended it on results, not on a promise of cheapness. The companies that flatten and last will be the ones that sold their boards the real trade, and could still point to the numbers when the first invoice failed to shrink.
Sources
- Martela and Nandram (2025). Buurtzorg: scaling up an organization with hundreds of self-managing teams but no middle managers. Buurtzorg ran near 8% overhead versus a ~25% sector average yet at a higher cost per hour (~EUR 54 vs 49), the core evidence that flat lowers overhead without lowering unit cost; overhead figure dates to the mid-2010s and headcount to early 2021. https://link.springer.com/article/10.1007/s41469-024-00184-y
- Gallup (2025). State of the Global Workplace 2025. Disengagement costs an estimated $10 trillion and manager engagement fell from 27% to 22%, the losses a flat, engaged structure buys down off the overhead line. https://www.gallup.com/workplace/659610/state-global-workplace.aspx
- Deloitte (2025). 2025 Global Human Capital Trends. Managers spend ~40% of their time firefighting, showing much of a hierarchy's cost is friction rather than salary. https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2025.html