Strip a manager's job down to what a flat organization still needs, and a surprising thing survives. It is not the approvals, the status meetings, or the control over who knows what. Those are the functions AI and self-managing teams absorb most readily. What remains is the part managers currently spend the least time on: developing the people around them.
Deloitte's 2025 survey of nearly 10,000 leaders found managers giving only 13 percent of their time to developing their people, against nearly 40 percent lost to firefighting and administration, with 36 percent saying they felt unprepared for the people-management side of the role in the first place (Deloitte, 2025). Read that as a forecast rather than a complaint. The 40 percent is the part flattening removes. The 13 percent is the part it leaves behind, and most managers have almost no practice at it.
The 13 percent problem
The reason the coaching function survives is that it is the one part of management that resists both automation and distribution. A self-managing team can set its own roster and make its own calls, but it still contains a person who is stuck, a conflict that needs a third party, a skill that has to be built. That is developmental work, specific to individuals, and it does not compress into a policy or a model. It is also, on Deloitte's numbers, the work managers were doing least even before anyone talked about flattening (Deloitte, 2025).
This is the quiet difficulty in the whole transition. The story of the great flattening is usually told as subtraction, as if removing the controlling manager simply leaves a leaner team. But the residue is not nothing. It is coaching, and coaching is a craft. Telling a manager whose week was 40 percent firefighting that the firefighting is gone and coaching is now the job is not a relief. For many it is being asked to do the part they were worst at and had the least time to learn.
The coach without a badge
The organization that has industrialized this separation is Buurtzorg. It supports more than 900 self-managing teams and roughly 14,700 staff with a group of coaches who hold no formal authority and carry no responsibility for team productivity (Martela and Nandram, 2025). That design is the point. By stripping control out of the developmental role, Buurtzorg made it possible for one coach to serve dozens of teams and made it safe for teams to actually use them. A coach who cannot punish you is a coach you can be honest with.
The lesson is not that every company should adopt Buurtzorg's exact structure, which rests on specific conditions the authors are careful to name, including low interdependence between teams and highly motivated frontline staff (Martela and Nandram, 2025). The lesson is the separation itself. Control and development have been bundled into one role for so long that most organizations assume they are inseparable. Buurtzorg shows they come apart cleanly, and that development scales far better once it is no longer weighed down by the accountability and authority that control demands.
The bundling also explains why so many managers were poor coaches. When one person holds both the developmental conversation and the performance verdict, the people being coached learn to manage the relationship rather than use it. Honesty becomes risky, because anything admitted can later count against you. Separate the two, as Buurtzorg does by giving coaches no say over productivity, and the developmental conversation can finally be what it claims to be. That is not a soft benefit. It is the difference between a team member who hides a problem until it is a crisis and one who raises it while it is still cheap to fix.
Rebuilding the job around the residue
Gallup's data suggests the payoff is real when the redesign is done well. In best-practice organizations, 79 percent of managers are engaged, nearly four times the global average of 22 percent, a gap Gallup attributes to how the role is designed rather than to the economy (Gallup, 2025). A role built around development, rather than around firefighting a manager never wanted, is a role people can sustain. The engagement number is not a reward for good intentions. It is what a well-scoped job produces.
The practical implication runs against the grain of most flattening programs. Cutting the controlling manager is the easy half, and companies are doing it at speed, with US middle-management postings down 42 percent from their 2022 peak (Deloitte, 2025). Building the coaching capability that has to remain is the hard half, and almost no one is investing in it at the same pace. Ricardo Semler's model of self-management always assumed that leaders would spend their time developing capable people rather than controlling them. The flattening is finally forcing that assumption on everyone else. The companies that thrive will be the ones that treat the surviving 13 percent not as a leftover but as the whole job, and teach their managers to do it before the firefighting they used to hide behind is gone.
Sources
- Deloitte (2025). 2025 Global Human Capital Trends. Managers spend only 13% of their time developing people and 36% feel unprepared to manage people, showing the surviving role is the one they practice least. https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2025.html
- Martela and Nandram (2025). Buurtzorg: scaling up an organization with hundreds of self-managing teams but no middle managers. Buurtzorg's coaches hold no formal authority and no productivity responsibility, showing development can be separated cleanly from control. https://link.springer.com/article/10.1007/s41469-024-00184-y
- Gallup (2025). State of the Global Workplace 2025. Best-practice firms reach 79% manager engagement versus a 22% average, evidence that a well-designed role built around development is sustainable. https://www.gallup.com/workplace/659610/state-global-workplace.aspx