Spans of Control

The Economist's Boss Class podcast recently interviewed Robin Dunbar of "Dunbar's number" fame, which reminded me about some other reading I'd done about spans of control. How large can a team get before it becomes ineffective? Surprisingly small!

I've written a few items under the "philosophy" tag, including:

1. One-on-one meetings
2. Outrage - uses and defuses
3. Scar tissue
4. Leader standard work
5. Communications channels
6. Remote work
7. e-Commerce
8. Management by Trust book review
9. Building for iteration
10. Temporary team leadership

Dunbar's Number

The discussion on The Economist's podcast reminded me about some thoughts on the best size for an organization. Robin Dunbar researched social evolution in primates and found that humans can have a maximum of 150 "meaningful relationships". This seems to be a function of brain size and holds stable across all cultures.

In a corporate organization, this means that relying on social relationships can be effective in small organizations, but very quickly the organization will outgrow its employees' ability to maintain relationships. At that point, formalized processes have to take over and organizational flexibility may suffer.

If a limit of 150 people isn't small enough, Dunbar goes on in the interview to state that the optimal size for a conversation (and therefore a meeting and/or team that needs to actually solve a problem) is only four! Any larger, and any single conversation almost always splits into two separate conversations:

...that sets this up a limit on the number of people you have in a conversation and indeed the number of people Shakespeare puts on the stage having a conversation together in any given scene. This conversational limit is so strong you can just see it work every day if you go to a reception and just watch people talking to each other it’s absolutely rigid. Four people in a conversation it’s fine. Five people in the conversation within 30 seconds it will split. [emphasis added]

Span of Control

Optimal team size is also remarkably low. When doing some research on infantry squads several years ago (for some detail on a BattleTech campaign), I ran across Major Paul E. Melody's monograph "The Infantry Rifle Squad: Size Is Not the Only Problem". In the aftermath of World War II, the US Army had data and experience on a vast scale of what had worked and what hadn't for the organization. Looking toward the future, Army leadership commissioned several reports and studies to document those lessons learned. Subsequent wars and experiences (e.g. Korea, Vietnam) led to additional studies.

These studies were not directly equivalent to managing non-combat teams: most notably, the size of teams in industry normally don't need to worry about casualties like army squads ("attrition"), but in general the ideal number of people for a squad leader to control was around five: "It was harder to control seven or more [including the squad leader] than it was to control five or less." (page 33) Because of attrition, the post-WWII recommendation for squad size was around nine people, including an assistant squad leader to help spread the management load.

Something that I particularly liked about this monograph was that it also pointed out that the rifle squad had begun to suffer from a lack of rifles: with the proliferation of special weapons in the squad (grenade launchers, machine guns, etc.), too few soldiers were available for close combat and the strenuous tasks better-suited for people carrying lighter equipment. It's worth remembering that organizations need plenty of people to do the "basic" work, and the people with the flashy tools only exist to support them.

Conclusions

Teams and organizations thrive at sizes much smaller than we might expect, and certainly smaller than many teams that I've seen in major organizations.

We can decide to decrease our spans of control by limiting managers to six or fewer direct reports, or maybe even three (to match conversation size). That may add layers to the organization, so we need to balance individual managers' span of control with keeping a reasonably "flat" organization. In addition, when possible the total number of people involved in a particular venture (e.g. supporting a project or product line) might be best limited to significantly fewer than 150: informal relationships can provide far better results and drive innovation in the way that disconnected silos can't.

What's your ideal team size? Have you been in teams that have ended up too big? Too small? Let me know at blog@saprobst.com or in the comments below!