Why Organizations Need Structure

The part overview promised that a big organization is a machine you can learn to read. Before we compare the different shapes that machine can take — functional, divisional, matrix — we need to understand the force that creates those shapes in the first place. If you skip this, org charts look arbitrary, like decorations someone drew on a whiteboard. They aren’t. They’re a response to a very specific pressure that shows up the moment a group of people gets large enough.

This page gives you that pressure in plain terms, and one idea you’ll carry through the rest of the part: structure is not bureaucracy for its own sake — it’s how a group stays coordinated once it’s too big to coordinate by talking. Get this, and every org chart you ever see becomes readable.

Start with a group that has no structure

Picture a coffee cart with three people: one takes orders, one makes drinks, one restocks and cleans. There’s no org chart. There doesn’t need to be. If something goes wrong — the milk runs out, a customer is upset, the card reader freezes — whoever notices just turns to the other two and says it out loud. Everyone hears everything. Coordination is free because the whole team fits inside one conversation.

This is the natural state of small groups, and it works remarkably well. A three-person startup, a small clinic’s morning shift, a two-cook kitchen — none of them need process documents. They coordinate the way friends plan dinner: informally, in the moment, by talking.

So the real question of this whole part isn’t “why do organizations have structure?” It’s “why can’t they stay that simple?” The answer is math.

The principle: connections grow faster than people

Here’s the force that changes everything. When you add people to a group, you don’t just add workers — you add relationships. And relationships grow much faster than headcount.

With 2 people there is 1 connection between them. With 3 people there are 3. With 4 people there are 6. The pattern is that each person can connect to every other person, so the number of possible two-person links is roughly n × (n − 1) ÷ 2 — which, for large teams, grows in proportion to n squared. Double the people and you nearly quadruple the possible connections.

People (n)     Possible one-to-one connections
    3                        3
    5                       10
   10                       45
   20                      190
   50                    1,225
  150                   11,175

Read that table slowly, because it’s the entire reason the rest of this part exists. Going from 10 people to 50 — a fivefold increase in headcount — creates a twenty-seven-fold increase in the number of pairs who might need to talk. A group that grew 5× in size grew almost 30× in coordination cost.

The coffee cart worked because 3 people meant 3 connections, all of them fitting inside one conversation. At 50 people, there are over a thousand possible one-to-one links. Nobody can hold a thousand relationships in their head. The “just say it out loud and everyone hears” model doesn’t slow down — it shatters. Messages get missed. Two people solve the same problem twice while a third problem goes untouched. Decisions get made and then quietly unmade by someone who never heard them.

The evidence — the limits of informal coordination

There’s a well-known idea from anthropologist Robin Dunbar that humans can maintain only around 150 stable relationships — the rough size, historically, of a village or a tight-knit company before it needs formal systems. The exact number is debated and shouldn’t be taken as precise. But the underlying point is robust and matches what managers see: informal, everyone-talks-to-everyone coordination has a ceiling. Somewhere in the low hundreds of people, and often well before, a group can no longer run on relationships and hallway conversation alone. That ceiling is where structure stops being optional.

What structure actually is

So structure is the response. Here is the cleanest way to define it:

Structure is a deliberate reduction of who-talks-to-whom into a small number of predictable channels.

Instead of everyone potentially coordinating with everyone — the thousand-link chaos — structure says: you talk to these specific people, about these specific things, in these specific ways. A warehouse picker doesn’t negotiate directly with the finance department; they talk to their shift lead, who talks to the operations manager, who talks to finance. Most of the thousand possible links are deliberately not used. That pruning is the whole point.

Think of it like a city. In a village, you can walk anywhere directly. In a city of millions, you don’t — you use roads, intersections, and traffic rules that channel movement into predictable paths. Nobody experiences the traffic rules as freedom, but without them the city would gridlock instantly. Structure is a company’s traffic system: fewer routes, but ones that actually move.

The common trap — treating structure as red tape

The most common mistake, especially from smart people who joined a company when it was small, is to see structure as pointless bureaucracy strangling the “good old days” when you could just walk over and fix things. Sometimes it is overgrown — we’ll cover that in Why Process and Bureaucracy Exist. But the instinct “let’s just remove the structure and talk directly” ignores the math: at scale, direct-to-everyone isn’t a nice option that structure took away — it’s a physical impossibility. The right question is never “structure or no structure?” It’s “does this structure fit the work we actually do?”

The two things every structure must divide

Every organizational structure, no matter how it’s drawn, is solving two problems at once. Naming them gives you a lens for the entire part.

1. The division of labor — who does what

You can’t have everyone do everything, so you split the work into parts and assign them: this team owns billing, that team owns delivery, this nurse handles intake, that surgeon operates. Specialization makes each part faster and better. A dedicated billing team gets sharp at billing.

2. Coordination — how the pieces fit back together

But the moment you divide labor, you create a new problem: the pieces have to recombine into a working whole. The billing team’s decisions affect the delivery team’s promises. The intake nurse’s notes have to reach the surgeon. Splitting work always creates seams, and coordination is everything a company does to keep those seams from leaking.

DIVIDE the work  ─────►  specialists move fast
      │
      ▼ (but now the pieces are separate)
RECOMBINE the work ───►  coordination keeps them in sync

Here’s the insight that ties it together: the way you divide labor determines how hard coordination becomes. Group people by function (all the engineers together, all the salespeople together) and coordination within each skill is easy but coordination across them — to ship one product — gets hard. Group them by product instead and the opposite happens. Every org chart in the next few pages is just a different answer to “where do we put the seams, and who stitches them?”

In practice — the seam you can hear

Sit in on almost any cross-team meeting and you’ll hear the seam directly. “We assumed you’d handle that.” “That’s not our team’s job.” “Nobody told us the date moved.” None of these are people being lazy or difficult. They’re the sound of a coordination seam — a place where divided labor didn’t fully recombine. Once you hear meetings this way, you stop blaming individuals and start asking a structural question: which channel was supposed to carry that information, and why didn’t it?

The trade-off that runs through everything

Now the single most important idea in this part, and the one to hold onto:

Every structure optimizes some things at the expense of others. There is no structure that is good at everything.

This follows directly from the two divisions. Any way you group people makes coordination easy in one direction and hard in another — you can’t make it easy in all directions at once, because that would mean everyone coordinating with everyone, which is exactly the chaos structure exists to prevent. Choosing a structure is choosing which coordination you want to be cheap and which you’re willing to make expensive.

A functional structure makes a company excellent at deep skill but slow to ship across teams. A divisional structure makes each product fast but duplicates effort and can let the whole company drift apart. A flat structure keeps decisions fast but overwhelms the few people at the top. None of these is “the right answer.” Each is a bet about what matters most. When you evaluate an org chart — including your own company’s — the sharp question is never “is this good?” but “good at what, and paying for it where?”

The chart is not the whole story

One last distinction, because it’s where most beginners go wrong. There are actually two organizations layered on top of each other.

• The formal organization is the chart — the official boxes, reporting lines, and titles. It tells you who is accountable for what and who decides what.
• The informal organization is how work actually flows — who people really go to for answers, who gets consulted before a decision “official” or not, whose blessing a project quietly needs, which two people from different departments trust each other and route around the process to get things done.

Both are real, and both matter. The formal org tells you how the company is supposed to work; the informal org often tells you how it really works. A new manager who reads only the chart will be baffled when the person with the impressive title has no real pull, while a quiet coordinator with a modest title turns out to be the hinge the whole department swings on.

Case study — the value of the informal network

When companies went fully remote almost overnight in early 2020, many discovered how much they had been relying on the informal organization without naming it — the hallway question, the “grab you for a second” clarification, the lunch-table trust between two departments. The formal structure (the chart, the meetings) survived the move to video calls largely intact. The informal structure did not, and a lot of coordination that used to happen invisibly suddenly had to be made explicit. The lasting lesson: the informal org is doing real load-bearing work, and you often only see how much when it breaks. Treat the specifics as illustrative, but the pattern held across many organizations.

We’ll come back to the informal org directly in How Structure Shapes Communication. For now, just hold both maps in your head at once. Reading only one of them is how good people misjudge organizations.

Try this

Pick one organization you know well — your workplace, a club, a place you’ve volunteered. On a single sheet of paper, draw two maps side by side. On the left, the formal org: the official boxes and reporting lines as HR would draw them. On the right, the informal org: draw a line between any two people who actually talk to get real work done, regardless of the chart. Then circle the three biggest differences between the two maps. Those differences are where the structure and reality have drifted apart — and they’re usually where either the best coordination or the worst friction in that organization lives.

Reflect

1. Think of a group you were in that grew from small to large. What specifically stopped working — and can you connect it to connections outgrowing headcount?
2. In your current organization, where is coordination cheap (things just happen) and where is it expensive (endless meetings, dropped balls)? What does that reveal about how labor is divided?
3. Name one thing your organization’s structure is clearly good at. Now name the thing it’s paying for that with. Did you already know the second one?
4. Who in your organization has more real influence than their title suggests — and who has less? How did you figure that out?
5. Recall a recurring frustration at work you’d normally blame on a person. Could it actually be a coordination seam — a place divided labor failed to recombine?

Show reflections

1. The useful move is naming the specific failure — messages missed, duplicated work, decisions unmade — and seeing it not as people getting worse but as the group crossing the point where informal coordination stops scaling. That reframe is the whole page in one memory.
2. Cheap coordination usually means those people share a boss or a team (labor divided so the seam is internal); expensive coordination usually means the work crosses a seam between two groups. You’re reverse-engineering the structure from where the friction lives.
3. A strong answer states the trade-off as a pair — “fast to ship but teams duplicate work,” “deep expertise but slow across functions.” If you can only name the good half, you’ve found a blind spot worth investigating before it costs you.
4. This is you reading the informal org. Good answers usually cite behavior, not title — who gets consulted, who others check with, whose objection actually stops things. Noticing you inferred it from behavior is the skill.
5. The point is to practice hearing frustration as a structural signal rather than a character flaw. Not every problem is structural, but many “unreliable person” complaints are really “the channel that was supposed to carry this didn’t exist” — and that’s fixable in a way blame never is.