The flattening was sold as a removal of bureaucracy and it delivered exactly that. A layer came out, decisions travelled a shorter distance, and the reporting line from the work to someone who could authorise things got usefully shorter.
What it also did, arithmetically and unavoidably, was widen every remaining manager's span. The people who used to run six now run sixteen, and the organisations that did it are working out what that costs.
The number that was never chosen
Span of control was, for most of the last decade, an outcome rather than a decision. Nobody set it. It emerged from headcount divided by whichever managers survived the last reorganisation, and it drifted upward every time the company grew without adding management.
Companies now putting a target on it are landing in a narrower range than the current reality, and the reasoning is consistent. Beyond a certain number, a manager can still run the work — standups happen, tickets close, the quarter lands — but stops doing the parts of the job that are not the work: development conversations, honest performance assessment, noticing that someone is about to leave.
Those are precisely the activities with delayed consequences, which is why the cost took years to appear and why it appears as something else. Attrition in a team of eighteen reads as a market problem. Weak internal candidates for a senior opening read as a hiring problem. Neither gets traced back to the span decision that made the development conversation impossible to hold.
The evidence organisations cite internally tends to be their own exit data. When people leaving are asked what would have changed their mind, the answers cluster around things a manager with time would have caught early — a clear path, a conversation about scope, a problem raised before it calcified. That is also why internal mobility is the cheapest recruiting channel most companies fail to use: moving someone requires a manager who knows them well enough to sponsor the move.
Setting a span is easy and staffing it is not, because the layer that was removed is the one that has to come back. That is the reckoning companies cutting middle management are now working through in the other direction, and it is more expensive on the way back in than it was on the way out, since the people who would have grown into those roles left during the years when nobody could develop them.
The organisations doing it credibly separate two things the flattening conflated. Layers of hierarchy — how many approvals a decision passes through — genuinely should be few. Span — how many people one person is responsible for developing — is a different variable, and treating them as the same number is what produced sixteen.
It shows up in succession too, and later than anywhere else. A manager who cannot hold development conversations does not produce a deputy, and a function without a deputy has a bench exactly one name deep — which nobody notices until the name resigns.
The uncomfortable part is that the correct span depends on the work, so no benchmark settles it. A team doing well-understood repeatable work supports a much wider span than one doing ambiguous work with junior people in it, and most companies applying a single target across both are about to learn that at some expense.
