Hierarchy, Span of Control, and Layers

Three Ways to Draw the Org Chart showed you the shape of an organization from the side — how it’s carved into functions, divisions, or a grid. This page turns the chart on its edge and looks at it from top to bottom. Why does a company have levels at all? Why does your boss have a boss, who has a boss? And why do some organizations feel fast and close to the ground while others feel like shouting down a long hallway?

The answer comes down to three linked ideas: hierarchy (who has the right to decide what), span of control (how many people report to one manager), and layers (how many levels stack between the front line and the top). Get these three straight and a lot of otherwise mysterious things — why decisions crawl, why reorgs keep happening, why the same idea feels different in a 12-person shop and a 12,000-person one — suddenly make sense.

Hierarchy is a chain of decision rights, not a status ladder

The word “hierarchy” carries baggage. People hear it and picture a status ladder — corner offices at the top, cubicles at the bottom, and a scramble to climb. That picture isn’t wrong about how hierarchy often feels, but it badly misdescribes what hierarchy is for.

Strip away the status and here is the useful definition: a hierarchy is a chain of decision rights and accountability. For any given decision, someone has the right to make the call, and someone is answerable if it goes wrong. Hierarchy is simply the map of who those someones are.

Consider a hospital ward. A nurse can decide many things at the bedside on their own. Some decisions they escalate to the charge nurse. Some the charge nurse escalates to the attending physician. And a few — closing a ward, a major policy change — go higher still. Nobody thinks this is about who is a “better person.” It’s about matching each decision to the person with the right information, authority, and accountability to own it. That’s hierarchy doing its actual job.

Decision                          Who decides        Why here
--------------------------------  -----------------  ----------------------------
Which patient to see next         Nurse              Closest to the situation
Reassigning staff this shift      Charge nurse       Sees the whole ward
Changing a treatment protocol     Attending doctor   Has the clinical authority
Closing beds for renovation       Ward manager       Owns budget + consequences

Read top to bottom and you see the real logic: decisions rise to the level where someone has enough scope and accountability to own them, and no higher. A healthy hierarchy pushes each decision to the lowest level that can responsibly hold it. (Who “should” hold which decision is the whole subject of the next page, Centralization vs. Decentralization.)

In practice — the two questions a clean hierarchy answers

When a hierarchy is working, any person can answer two questions instantly for any decision they face: Can I decide this myself? and If not, who can? When people can’t answer those — when they don’t know whether they’re allowed to act, or who to ask — that’s not a personality problem, it’s a broken chain of decision rights. Most “we’re so slow” complaints trace back to those two questions having fuzzy answers.

Span of control: how many people one manager holds

Now zoom in on a single manager. Span of control is just the count of people who report directly to them. A team lead with four direct reports has a span of four. A regional director with fifteen store managers under them has a span of fifteen.

That single number carries a real trade-off, and it’s the heart of this page. A manager has a finite amount of attention, time, and energy. Divide it among more people and each person gets less.

• A narrow span (say, 3–5 reports) means the manager can give each person real attention: frequent one-on-ones, close coaching, fast answers. But narrow spans need more managers to cover the same number of workers — which, as we’ll see, stacks up more layers.
• A wide span (say, 10–20+ reports) means fewer managers and fewer layers. But each person gets a thinner slice of the manager’s attention, so wide spans only work when people are experienced, the work is similar and predictable, and they don’t need much day-to-day direction.

So the honest way to state it is not “wide good, narrow bad” or the reverse. It’s a trade: span buys you flatness at the cost of attention per person. The right span depends on the work.

When a WIDE span works well          When a NARROW span is needed
-----------------------------------  -----------------------------------
Work is routine and repeatable       Work is complex or novel
People are experienced, need little  People are new, learning, or in a
  direction                            high-stakes/high-judgment role
Tasks are similar across the team    Each person's work is very different
Little coordination between people   Heavy coaching or coordination needed

A shift supervisor in a busy cafe can reasonably oversee twelve baristas: the work is well-defined, the team is trained, and problems are visible on the floor in real time. A lead surgeon mentoring three surgical residents on delicate cases cannot oversee twelve — the work is high-stakes and each case needs deep attention. Same word, “span of control,” but the right number is wildly different because the work is different.

The evidence — there is no magic span number

For a century people have hunted for the “right” span of control, and older rules of thumb often landed around five to seven reports. Modern research and practice are much less tidy: studies generally find the workable span depends far more on the nature of the work and the experience of the people than on any universal number. Routine, standardized work supports much wider spans; complex, interdependent, or novel work forces narrower ones. Treat any single “ideal number” you hear with suspicion — the useful question is never “what’s the right span?” but “what’s the right span for this work, with these people?”

The arithmetic: span drives layers

Here’s where the two ideas connect, and it’s just arithmetic. If you know how many people an organization has and roughly what its average span of control is, you can predict how many layers it needs — because each layer can only manage so many people below it before it needs another layer to help.

Picture an organization of about 4,000 front-line workers. Watch what the span does to the number of layers:

Average span of ~5                    Average span of ~15
------------------------------------  ------------------------------------
1        CEO                          1        CEO
5        run ~4,000 below             15       run ~4,000 below
25       ...                          225      front line
125
625
~3,125 front line
= about 6 layers                      = about 3 layers

Same 4,000 people. Narrow the average span and you get roughly six levels between the front line and the top; widen it and you get about three. Span and layers are two views of the same choice. You cannot independently pick “give every manager only a few reports” and “keep the org flat” — the math won’t let you. Choosing one chooses the other.

This is not an abstract point. Those extra layers have real consequences for how the whole organization behaves.

Why more layers slows and distorts information

A layer is a relay point. Every piece of information — a customer complaint, a market signal, a decision request, a change in strategy — has to pass through each layer on its way up or down. Two bad things happen at every relay.

It gets slower. Each layer is a queue. The message waits for someone to read it, react, reword it, and pass it on. A request that could travel from front line to decision-maker in one hop through three layers now waits three times, in three inboxes, behind three other priorities. In a fast-moving situation, the delay itself can be the failure.

It gets distorted. This is the telephone game — the children’s party game where a whispered sentence passes down a line of people and comes out the other end as nonsense. Every retelling drops a nuance, adds an assumption, or softens an awkward truth. By the fourth or fifth relay, the message that reaches the top can be meaningfully different from what the front line actually said.

Front line:  "Customers are furious about the new checkout —
              it's costing us sales, this week."
   ↓ layer 1: "Some customers are unhappy with checkout changes."
   ↓ layer 2: "There's been feedback on the checkout update."
   ↓ layer 3: "The checkout project is getting attention."
Top hears:   "Checkout is going fine, people are engaged with it."

The urgency, the anger, and the sales impact all evaporated — not because anyone lied, but because each relay smoothed the edges a little. This is why leaders in tall organizations are so often the last to hear bad news, and hear it faded. It’s also why layers dilute accountability: with more relays between the decision and its outcome, it gets harder to say who actually owns the result.

Case study — Amazon’s six-page memo

Amazon is well known for a practice it adopted in the early 2000s: instead of slide decks, meetings open with everyone silently reading a densely written six-page narrative memo. One reason it works is that a written narrative resists the telephone-game decay of information passing up through layers — the reader gets the author’s actual reasoning, caveats and all, not a summary of a summary of a summary. The broader lesson isn’t “ban slides”; it’s that in any layered organization, how information travels between levels matters as much as the levels themselves. Design the channel, not just the chart. (More on this in How Structure Shapes Communication.)

Tall vs. flat: two ways an organization can feel

Put it together and you get two archetypes, each with honest costs and benefits.

A tall organization has many layers and narrower spans. A flat organization has few layers and wider spans.

                TALL                          FLAT
             (many layers)                 (few layers)

Attention    High — small teams,           Lower — big teams, less
per person   close supervision              individual coaching

Speed of     Slow — many relays up          Fast — short path to a
decisions    and down                       decision-maker

Info         Distorted, delayed             Cleaner, faster, but
quality      (telephone game)               managers can miss detail

Career       Many rungs to climb;           Few rungs; growth is
ladder       clear promotions               sideways / scope, not title

Cost         Higher — you pay for           Lower — fewer manager
             lots of managers               salaries

Feels like   Structured, safe, but          Autonomous, fast, but
             bureaucratic and distant       can feel unsupported

Neither wins outright. A nuclear plant or a hospital wants some tallness — close supervision and clear escalation are features, not bugs, when mistakes are catastrophic. A small startup or a creative studio wants flatness — speed and autonomy matter more than tight oversight. The mistake is copying a shape that suited someone else’s work. The right shape falls out of your work’s real needs for attention, speed, and safety.

The common trap — flat as a slogan

“We’re a flat organization” is a popular boast, and it often hides a mess. Flatness doesn’t delete the need for decision rights — it just makes them implicit, which usually means the loudest or most senior person quietly decides everything while the org chart pretends otherwise. Real flatness requires people to be experienced enough to self-direct and clear enough about who owns what to not need a boss to referee. If your “flat” team is actually one overloaded manager with fifteen confused reports and no clear decision rights, you don’t have a flat organization — you have an under-managed one. Flatness is earned with clarity, not declared by removing titles.

Why layers accumulate — and why “delayering” keeps happening

If layers slow and distort things, why do organizations keep growing more of them? Because layers accrete quietly, one reasonable-seeming decision at a time.

A manager’s span gets uncomfortably wide, so the company promotes someone to a new lead role and splits the team — one more layer. A valued senior person needs a promotion to be retained, and the only path is “manager,” so a management layer is created to hold a title — one more layer. A new region opens and gets its own director between the existing structure and headquarters — one more layer. Each move is defensible on its own day. Nobody ever decides “let’s become slow and distorted.” The tallness emerges, unplanned, from years of individually sensible additions.

Then, every few years, someone at the top notices the organization has become slow, expensive, and out of touch — and launches a delayering effort: removing management levels, widening spans, pushing decisions down. It’s a recurring corporate ritual because the underlying pressure is recurring. Growth pushes layers up (more people need more coordination); the desire for speed and lower cost pushes layers down. The org breathes in and out between the two.

The evidence — the flattening trend

Since roughly the 1990s, many large companies have deliberately cut middle-management layers, driven partly by cost pressure and partly by better information tools — email, dashboards, and now shared software let a message reach many people directly without a chain of human relays. Studies of large firms generally find spans have widened and the distance between the top and the front line has shrunk over the decades. The lesson to take is directional, not numerical: the reason layers exist — to relay information and coordinate people — weakens when technology carries some of that load, which is why the pressure to flatten keeps returning.

The common trap — cutting layers without moving the decisions

Delayering fails when it’s treated as a headcount exercise. Removing a management layer while leaving all the old approvals and decision rights in place doesn’t make the org faster — it just gives the survivors more people to manage and the same slow decision paths, so everyone burns out and nothing speeds up. Flattening only works if the decisions move down with the removed layer: the people who lost a boss must gain the authority that boss used to hold. Cut the layer, keep the bottleneck, and you get the cost of flat with the speed of tall — the worst of both.

Putting it together

Three ideas, one machine. Hierarchy decides where each decision lives. Span of control sets how many people each manager can genuinely attend to, given the work. And layers — the direct arithmetic consequence of span — quietly govern how fast information moves, how much it distorts, how much the whole thing costs, and how it feels to work inside it. When you next feel an organization is slow, distant, or oddly hard to get a decision from, don’t reach first for “bad people.” Count the layers, check the spans, and ask where the decision rights actually live. The shape usually explains the feeling.

Try this

This week, sketch the chain above your own seat — you, your manager, their manager, and so on up as far as you can name, to the top. Count the layers between the front line and the highest person you can name. Then pick one recent decision that felt slow and trace its path: how many layers did the request cross going up, and the answer coming down? Finally, estimate your own manager’s span of control (how many people report to them) and ask whether the work you do needs the close attention of a narrow span or the autonomy of a wide one — and whether you’re actually getting the one it needs.

Reflect

1. In your organization, for a typical decision you face, can you instantly answer “can I decide this myself, and if not, who can?” Where is that chain of decision rights clear, and where is it fuzzy?
2. What is your manager’s span of control, and does the nature of your work call for a narrow span (close attention) or a wide one (autonomy)? Is there a mismatch?
3. Recall a time information reached the top distorted or too late. Which layers softened, delayed, or lost the real message — and what did that cost?
4. Is your organization more tall or more flat, and does that shape actually fit the work you do — or is it copied from somewhere the work was different?
5. Have you seen layers quietly accumulate, or a delayering effort come through? Did the decisions move when the layers did, or did only the boxes on the chart change?

Show reflections

1. This tests whether hierarchy is doing its real job for you. Clear answers mean the decision-rights chain is healthy; fuzzy ones (“I’m not sure if I’m allowed” or “no idea who owns this”) are where speed leaks out. Notice that the fix is rarely a new person — it’s making the existing chain explicit.
2. There’s no right number, only a right fit. A wide span with novel, high-stakes work usually means you’re under-attended; a narrow span with routine work you’ve mastered can feel like over-supervision. Naming the mismatch tells you what to ask for — more autonomy, or more of your manager’s time.
3. The goal is to connect the abstract telephone-game idea to a concrete cost — a missed warning, a slow reaction, a decision made on a faded version of the truth. That cost is the real price of layers, and seeing it makes the case for shorter paths or better channels concrete.
4. A strong answer separates “what shape we have” from “what shape our work needs.” Tallness fits high-stakes, safety-critical, or highly regulated work; flatness fits fast, autonomous, lower-risk work. A mismatch — a flat org doing surgery-grade work, or a tall org doing creative work — is a real, nameable source of friction.
5. This is the recurring corporate cycle up close. If you’ve seen a delayering, the sharpest test is whether authority actually moved down with the removed layer. If the boxes vanished but the approvals stayed, the reorg cut cost without buying speed — the trap in the caution above.