Scott Kennedy

I have nine bosses, Bob: costs of org layering

November 12, 2022

I recently learned an old teammate of mine can now top Office Space’s Peter Gibbons by dropping the line “I have nine bosses, Bob”… and mean it literally.

How many bosses is too many? And what do they all do, anyway?

When I first started at Google, it was a 20,000 person company and my management chain looked something like:

Me -> Director -> VP -> VP -> SVP -> Larry

An extra layer appeared shortly after as we continued to grow, which felt appropriate.

Today Google is a 200,000 person company. If I were starting today it might look like:

Me -> Manager I -> Manager II -> Manager III -> Director -> Senior Director -> VP -> VP -> SVP -> Sundar

Nine bosses, Bob!

It turns out there is fairly consistent advice on how many direct reports a manager should have12. Some will point to Andy Grove in the 1980s classic High Output Management as the origin. Grove came up with the range 6-8 because:

Now, maybe you’ll argue with that advice. But Google took lots of lessons from Grove and I’m fairly sure they took this one3. Larry Page once mentioned that he thought 7 was a reasonable average. He was asked about hierarchy at TGIF and did the rough math below to justify a Googler having 5 bosses at 20,000 employees:

Org depth Max Company size
1 8
2 57
3 400
4 2,801
5 19,608
6 137,257
7 960,800
8 6,725,601
9 47,079,208
10 329,554,457

If you recall my friend with 9 bosses, she’s one manager away from an org that could theoretically manage the population of most countries on Earth.

So why did the 6-8 rule stop holding true in practice?

First, this math has some flaws:

But we’re still off by ~2 layers of management even if we assume 5 manager reports per manager. Our 9 bosses could still be handling 2.4M employees.

The bigger issue I saw was rooted in incentives. Manager promotions are often4 earned by showing the ability to lead larger teams. Thus, a talented manager is often recruited or retained with promises of building large teams that require layers of management. They demonstrate their ability to run a large org by hiring in managers or layering existing teams ahead of the growth. In the extreme, a manager or executive who cares about this more than execution is un-lovingly referred to as an empire builder.

It’s no coincidence that my friend’s number of managers is almost exactly the number of promotion levels available on the management ladder5.

Are extra managers actually a bad thing? Was Grove right that the problem this creates is meddling? Sort of, yes.

I once heard a well respected executive describe the job as two primary duties; (1) set a compelling vision; and (2) hire a great team to execute it. While he was at Google, now CEO of Tailscale Avery Pennarun wrote a great essay on the view of a big company from the executive level that had similar conclusions6.

So what happens if you have a VP who reports to a VP who reports to a VP? The vision and leadership changes tend to cascade and multiply. Getting so many layers of people to understand strategy consistently is very hard. All of this churn results in project delays or cancellations, and frustration for individuals.

Solving this problem is very hard:

But there’s a far simpler solution: work at a small company.

My current employer, Replit, has fewer than 100 employees. The deepest management chain looks like this:

Individual Contributor -> Manager -> Executive -> Amjad

Given the simplicity of the organization, even the CEO can be knowledgable about details of key projects and decisions. Truth about what is working or not travels fast. Important changes in strategy or direction can be communicated by a single email to all employees.

I specifically chose a small company for this reason. But I didn’t anticipate just how many aspects of my job would be simplified. There are no re-alignment meetings. Decision making can be pushed more to individuals because consistent choices are more obvious. The biggest incentive for each individual is still make the company actually succeed.

As our engineering department grows above 50 people, we start to consider having an additional layer of management. This feels easier to navigate. The people stepping into management are doing it because they think it’s the most helpful thing they can do, not to climb a ladder. More people look up to the senior ICs than the managers career-wise and that feels healthy.

If Google still felt okay at 20,000, I’m pretty sure Replit can stay easy well beyond 100.

  1. If somebody insists on calling this a manager’s span of control, be wary of their advice as they have been reading too many academic studies of management.
  2. Former Navy SEAL Jocko Willink quotes 5-10 as the possible range in his book Extreme Ownership. Interesting to see the range be consistent across both tech and the military.
  3. After a famously failed 2002 “no managers” experiment.
  4. This was not actually a rule. I saw a talented leader promoted to Director with a small, specialized, and highly impactful team (and no layering). But having a big team is definitely the easier path.
  5. This number is not publicly shared, but I believe it is 10 (excluding whatever level Sundar is considered to be).
  6. Some of those thoughts made it into this excellent public essay.