Spans and layers: how many direct reports is too many?

Two numbers describe the shape of any company. Span is how many people report to a single manager. Layers are how many management levels sit between you and the person doing the work. Almost every “why are we so slow now?” problem in a 20–100 person company traces back to one of them.

Span: it's about the work, not a number

The instinct is to ask for a rule — five reports, seven, ten. But span follows the work. When the work is similar and repeatable, one manager can genuinely support a lot of people, because the decisions rhyme. When the work is complex and every call is a judgment, the same manager drowns at a fraction of that number, because every report needs real thinking, not a quick yes.

So the useful test isn't the headcount under a manager. It's this: is each person actually getting decisions made, getting coached, getting unblocked? Or has the manager quietly become a status-meeting host who hasn't moved anyone's work forward in weeks? A span is too wide the moment management stops happening — and that number is different on every team.

A span isn't too wide because of the count. It's too wide the moment people stop getting managed.

Layers: every level should own something

Layers are the opposite failure. Each management level you add puts another handoff between a question and an answer. A little of that is the price of size. Too much of it, and decisions die in transit — routed up, sideways, and back down before anyone who can say yes ever sees them.

The tell for a redundant layer is a manager who manages one or two people. That seat usually exists to relay information, not to own an outcome. It feels like structure, but it's friction wearing a title. Collapsing it almost always speeds the company up and gives the people below a real decision-maker instead of a messenger.

The two failures look identical from your chair

Here's why this is hard to self-diagnose: spans too wide and layers too deep produce the same symptom — things feel slow, balls get dropped, and you can't name a single person at fault. But the fixes are opposites. Too-wide spans need more structure: split the team, add a seat. Too-deep layers need less: collapse a level, widen the survivors. Guess wrong and you make it worse — you add a layer to an org that was already too tall, or you flatten a team that was already unmanaged.

That's why the shape has to be read, not assumed. You have to look at where decisions actually route — not where the org chart says they do — before you touch the structure.

Flat isn't the goal

“Flatten the org” gets treated like an unqualified good. It isn't. Flat reduces layers and speeds decisions, but push it too far and you get spans so wide that nobody is actually managed and your best people burn out covering the gap. The target isn't flat for its own sake. It's the fewest layers you can run while every single person still has a real manager and a clear owner for their outcomes.

How a Staff Analysis reads it

This is the Structure layer of a Staff Analysis. We map your actual spans and layers, find where decisions stall or pile up, and separate the seats that are too loaded from the ones that are only relaying. You get the few role-level moves that rebalance the shape — widen here, collapse there, split this seat — ranked, with a 90-day sequence. Role-level only: the work points at seats and decisions, never at named people.