5, 15, 50, 150: when to redraw the org chart

Series「People Don't Follow the Org Chart」 · Part 2 of 13 · Part 1 · How people group
Prologue：The people who aren't on the org chart

An org chart has to be redrawn at roughly 15, 50 and 150 people, and most Japan entities redraw it late. Here is how it usually goes. A Japan entity grows from three people to thirty, and the head still runs it the way he did at three. Every decision is his, and everyone comes straight to him. At thirty it starts to jam. At fifty it stops. Most people call this "the boss won't delegate." I'd put it more precisely. The chart needed redrawing, and nobody knew when.

A sequence of numbers

In the 1990s the anthropologist Robin Dunbar proposed that there is a ceiling on the number of stable social relationships a person can keep up, at around 150. That number has been quoted so often that people treat it as the whole finding. The more useful part of his work is the sequence. About 5 intimates, 15 people you can rely on, 50 you stay in regular contact with, 150 whose names and relationships you know. Each layer is roughly three times the one before.

The dispute deserves a plain statement. Dunbar's number comes from a correlation between neocortex size and group size in primates, extrapolated to humans. A 2021 reanalysis of the same data produced confidence intervals running from a handful to several hundred, and concluded that 150 has no statistical privilege. Supporters point out that these layers keep turning up in military units, village sizes and social network data.

So I don't treat 150 as a biological constant. I use the sequence as a set of rough thresholds from experience. Each marks a point where the way relationships are held together has to change. The exact numbers can drift.

At each step, the mechanism changes

Up to 5, you need no organization. Everyone knows everything, and the division of labor is unspoken.

Around 15 is the ceiling of one team. One person can still judge everyone's work directly, but who owns what has to be made explicit. This is the first redraw, from "all of us" to "a few roles."

Around 50, one person can no longer see everyone, and a middle layer appears. This is the second redraw, and the easiest to get wrong. Middle management is an added layer that filters information. Cutting the boss's job into pieces is the smaller part of it. The "good news only" pattern in my organizational diagnosis almost always shows up here.

Around 150, people stop knowing each other. An organization held together by relationships ends here. After this point it is held together by systems, grades and written rules. That is the third redraw, from people managing people to systems managing people.

Redraw the org chart one step before the threshold, not after the jam

Dunbar’s sequence (5, 15, 50, 150) used as rough thresholds from experience. The exact numbers can drift

1. Up to 5No organization neededEveryone knows everything; the division of labor is unspoken
2. About 15First redraw: all of us → rolesMake explicit who owns what. Define three or four roles with one owner each; the head manages only them
3. About 50Second redraw: add a middle layerNo one sees everyone, and the middle layer filters so only good news travels up. Before appointing, decide what may and may not be filtered, and write down the three kinds of news that always travel up
4. About 150Third redraw: people → systemsPeople stop knowing each other, and pay set by relationship blows up. Start writing grade and pay rules

Running a 60-person company the 15-person way is bigger than any one manager’s problem. Compare today’s headcount with the headcount your chart was drawn for.

The design decision: redraw one step early

Charts usually get redrawn after the jam, about six months late. Treat these numbers as tripwires and act before you reach them.

As you approach 15, define three or four roles, each with one owner, and have the head manage only those people directly. As you approach 50, before you appoint middle managers, decide what they may filter and what they may not, and write down the three kinds of news that must always travel up. As you approach 150, start writing grade and pay rules. Setting pay by relationship works until then. After that, it always blows up.

Japan adds one variable. Japanese employees have a strong sense of place (see the previous post). So a team under 15 holds together more tightly than its Western equivalent, but the 50 threshold is harder to cross, because adding a middle layer means cutting one place into several. Cut by business line, and avoid cutting by function, so each new place has complete work to do.

The ruler I use in a diagnosis

The patterns layer of my organizational diagnosis asks whether the management style matches the size, and this sequence is the ruler I use. A 60-person company run like a 15-person one is a bigger problem than any single manager. Whether two heads at the top can work, which I covered in a separate post, also depends on size.

How many people does your company have now? And how many did it have when the chart was drawn?

In this series
Previous：Why global matrix structures break in Japan
Next：The favor ledger is more honest than the org chart