You Removed the Management Layer. The Coordination Work Did Not Go With It.

The coordination surface in a flat organization is not smaller. It is larger, and undesigned.

When a management layer exists, coordination routes through it. Information that needs to cross team boundaries crosses them at the manager’s level. Decisions that require context from multiple functions get that context assembled in one place. The layer is not glamorous work. It is not the work anyone announces at an all-hands. But it is the work that makes the rest of the work possible. Remove the layer without reassigning the work, and the surface area that manager was covering has to be covered by someone. Usually it is covered by everyone, which is the same thing as no one.

Three years ago I walked into a flat organization that had just come through an acquisition. The management team that preceded me had left. What remained was a group of engineers who each knew their own domain precisely and had almost no architecture for the coordination that crossed those domains. Every function knew its part. Nobody had been designed to carry the surface between the parts. The result was not chaos, it was something slower and more expensive: every coordination moment became its own negotiation. Too many cooks. Every problem that crossed a boundary required everyone at the table because there was no defined role to carry it.

The fix was not a reorganization announcement. It was three deliberate promotions and one outside hire, sequenced against the coordination gaps. First, a DevOps manager promoted from within: the surface between development and infrastructure was the most expensive coordination problem the team had, and it needed a single point of ownership held by someone who already understood the environment. Then a QA manager also promoted from within: someone who already understood the team’s standards and could carry the coordination between engineering output and quality. Then a development manager hired from outside: the one domain where fresh judgment was more valuable than institutional continuity. Each role absorbed a specific coordination surface. The sequence was not accidental. The most expensive gap got closed first.

The diagnostic question before removing any management layer is not “can we afford this role?” It is: what is the nature of this role? If the role is primarily information relay and task tracking, the work it does is being automated. AI does that faster and without headcount cost. That role should be reconfigured, not eliminated, because the judgment function it should have contained still needs a home. If the role contains genuine management judgment, coordination across functions, and accountability for outcomes that no individual contributor can own, removing it moves that judgment somewhere the organization has not designed for it. That somewhere is usually the most senior individual contributors on the team, who absorb the work without a title, a budget, or a mandate. That is Heroics: organizational dependence on individual performance to cover what the system was never designed to carry.

Every company announcing a flat structure is telling a story that lands well in a board deck: we are leaner, we removed the coordination tax, AI handles what managers used to handle. The story is not wrong about the technology. It is wrong about the work. AI has not replaced the judgment function of management. It has replaced the relay function. The judgment function, the part that understands context, absorbs ambiguity, makes calls across competing priorities, and is accountable when something fails, is still human work. Announcing a flat structure without designing where that work goes is not efficiency. It is offloading. The coordination tax does not disappear. It gets redistributed to people who did not sign up to carry it, and who will eventually price that into their decision to stay.

The question to ask before the announcement: what is every coordination function this layer currently performs, who will own each one after the layer is gone, and how will we know when that ownership is failing? If the answer is “we’ll figure it out,” the structure is not flat. It is incomplete.

I write about structural leadership for technical leaders in high-stakes operating environments. If you want to identify where your biggest structural constraints are, the LeadershipOS™ Scorecard maps them: https://theleadershiposbook.com/scorecard

---

I write about structural leadership for technical leaders in high-stakes operating environments. If you want to see where your system is load-bearing on you personally, the LeadershipOS™ Scorecard maps it: https://theleadershiposbook.com/scorecard