10 ms·

Number of levels in the hierarchy seems like a bad way to determine this. I've worked in a company with 500 employees (and three levels) and lots of bureaucracy

Number of levels in the hierarchy seems like a bad way to determine this. I've worked in a company with 500 employees (and three levels) and lots of bureaucracy and a company with 100,000+ employees (~6-8 levels depending on your division) with relatively little bureaucracy (yes, on occasion, I ran in to bad cases of it, but it didn't effect my day-to-day work). It's simply not possible to have 100,000+ employees and less than 6 levels, unless you have managers with tons of employees. But it is possible to be in a division of a large company where your group has relative autonomy and are empowered to do what they need to do. In my case, there were about 20 of us inside of a larger 500 person group. 95% of decisions were made within the 20 person group (obviously this is an approximation). Another 4% were made at the 500 person group management level. 1% were made by company-wide executives. In other words, my boss was accountable for most of what we did. On occasion, he's need to go to his boss for something. And very rarely, his boss would need to escalate to higher ups (where there were another 3 or 4 levels to the CEO). This company was very well run, but there are much smaller companies with fewer levels in the hierarchy and lots more bureaucracy.

awinter-py 7y ago
would be interesting to see some AI-assisted version of holacracy replace middle managers -- you'd still have low-level team leads and high-level strategy people, but a lot of the informational glue and project updates could be taken over by machines you'd need to go into a project like that (1) believing that middle managers eventually suppress more value than they add and (2) you're okay operating in a random mix of cowboy team leads and inconsistent organizational glue between them. you'd set broad parameters like 'don't over-duplicate work', 'pick projects that are more or less on-topic with our company', 'get third party verification of plans and estimates'. Firing could be some mix of 'future needs', performance and random. Moving teams would happen based on an internal market of skills & personnel budget. might be chaos but it could be fun.
- Matticus_Rex 7y ago
  I've been really interested to watch what Holacracy actually does (sans AI) in healthy, agile companies. A lot of the management overhead gets cut (while keeping functional management of the company, though a lot of it becomes process rather than people), and a lot of the projected issues are just things someone can propose a solution for and get it enacted via Holacracy governance. Springest is a really interesting example. Very anti-bullshit culture, rigid focus on OKRs and efficiency, and yet they're extremely light-hearted and lack the political crap most similarly high-performing companies have. Maybe AI makes it better in the long run, but I'm a believer already.
Misdicorl 7y ago
> It's simply not possible to have 100,000+ employees and less than 6 levels, unless you have managers with tons of employees Math does not check out. 5 levels for 100k employees results in 10 reports per manager (assuming a top level with 10 people rather than 1, which is approximately correct given CEO/CTO/CIO/etc/etc/etc). Maybe 10 qualifies as 'tons' for you? Going down to 4 levels gives 18 reports per manager, which feels a little closer to 'tons'.
- dvirsky 7y ago
  10 direct reports on average is a lot. It might be okay at the highest levels where you mostly make strategic decisions, but it's too much for a manager that's involved in day to day processes and decisions.
- chadash 7y ago
  I think we agree in principle, just different understanding of how to count levels. Based on the article, a company with a boss and one employee is two levels ("With only two levels, a boss and those who report to the boss..."). So for 100,000 people, with average 10 employees per manager: 1 Big Boss (Level 1) 10 (Level 2) 100 (Level 3) 1000 (Level 4) 10000 (Level 5) 100000 (Level 6) That's six levels. In practice, some people will have more than 10 and some less, but unless the people under you are doing menial work, I think 10 is probably middle of the road.
  - Misdicorl 7y ago
    Fair enough if you want to include the one person level. It seems rare that any large company uses a structure like that though
  - Aperocky 7y ago
    The math stop checking out at level 4. You have level 5 managers managing 10 each, and you have level 4 managers managing 10 teams each? That stops checking out. The higher you go the more disciplined you have to be about grouping relating teams under the same umbrella for efficient execution. In fact, the levels don't matter as much as the autonomy at each level. The more autonomy, provided that the right people are hired and are focused on the goal, makes it better and less managerial burden going up the chain.
throwawaypa123 7y ago
the 6 layers isn't really true, in part due to line managers can manage 15-30 people. If everyone is doing the same job that is fairly rote a single line manager can handle a lot of people, reducing layers. Additional CEOs tend to have a lot of direct reports. Sample Example (0 Layer) CEO --> 10 Direct reports is typical for large publicly traded companies. (1 Layer) C-Level --> 6-8 (10,000 / 8 --> 1,250) (2 Layer) SVP / VP -- > 6-8 (1250 / 6 --> ~200) (3 Layer) Middle management (200/ 6 --> ~30) (4 Layer) Line managers (1-25)
- dragonwriter 7y ago
  > the 6 layers isn't really true, in part due to line managers can manage 15-30 people Yes it is, IMO; any layer above line managers over nonprofessional staff should have a span of control not greater than about 3-5 subordinate managers with full-depth (1 level less than the manager) organizations and a similar number of supporting staff that either have no reports of their own or supervise organizations about 2-3 levels less deep than the manager above them. Yes, lots of real organizations have broader spans of control, but lots of real organizations also are what the author describes as “immoral mazes”.
- dtnewman 7y ago
  From the article: > With only one level, there’s nothing to worry about. With only two levels, a boss and those who report to the boss, the boss has skin in the game, no boss causing problems for them, and not enough reason to reward bad outcomes. With three levels, there are middle managers in the second layer, so one should be wary. Based on this wording, it seems like the CEO and the people under the line managers count as levels (levels though maybe not levels of middle management). By that reasoning, this structure would indeed be 6 layers. In reality, it varies by division if you are in a large company. If you join at the bottom of the legal department at Google, you probably have fewer layers above you than if you are a junior dev working on android.
- spoondan 7y ago
  There may be situations where 15-30 direct reports is workable, but that is a very large number of reports. Academic studies generally settle on a maximum effective team size around seven, give or take a few[1]. In my own experience, I saw diminishing returns and reduced satisfaction in my management performance after ten direct reports. These issues resolved with a restructuring. But you don't have to take academic studies or anecdotes to heart. Some back-of-the-envelope calculations suggest there are severe challenges to effectively and healthfully directly manage 30 people. Take the typical 40 hour week[2]. With 30 direct reports, that gives a maximum of 40/30 = 80 minutes/week to devote to each individual employee. If the manager does a weekly or bi-weekly 30 one-on-one, that leaves about an hour on average to deal with any issues. Of course, not every employee has high priorities issues at the same time. But we've only considered the individual, one-on-one work of the manager. There is also the team-wide and cross-team work. And, it turns out, team size also dramatically increases the time and effort these other facets take. Consider the typical daily stand-up (blech), used to keep the team aligned. If the manager allots (and strictly enforces) five minute slots per team member, that meeting still takes 2.5 hours. People won't be happy with that, and it will cut into the managers limited time to actually make progress on individual and team issues. Even if the manager splits the meeting (to spare team members for a super long meeting), the manager needs to attend each one to stay abreast of the big picture. Likewise, we presume that if a team has 30 reports, it's because what they are working on is of critical business importance. There are more demands from sibling teams in the same org, more budget and staffer reviews with higher-ups, more hours spent with HR and recruiting. (Alternatively, the manager is empire-building. But that doesn't really help the calculation. The time that the manager would've spent on delivering business value is instead spent on politics.) Even if you think you can keep all these balls in the air, it's hard to not conclude there are incredibly tight windows here. And the whole thing is very fragile. If you fail to identify, understand, and address a critical issue, it can quickly infect other team members. [1] See, for example, https://knowledge.wharton.upenn.edu/article/is-your-team-too-big-too-small-whats-the-right-number-2/ https://knowledge.wharton.upenn.edu/article/is-your-team-too... [2] I know. We often work more than that. The exact number doesn't really change the story, especially since we're talking about retaining our physical and mental health.
alecbenzer 7y ago
Given the author admits: > Note that those outside the company, such as investors or regulators, seem like they should effectively count as a level under some circumstances, but not under others. I think there's obviously some room to fudge with what the exact boundaries of the "organization" are. So yeah, maybe a sufficiently-empowered department head functions more like a CEO/boss.
dragonwriter 7y ago
> Number of levels in the hierarchy seems like a bad way to determine this. It's incomplete as a measure of the problem, but it's definitely a valid measure of a key risk factor. The span of control (# of direct reports) of each supervisor in a direct chain of command from the employee to the top-level manager is also a measure (higher is worse), as is the “span of reporting” (the number of supervisors[1] to which each employee reports) from the employee up to the top-level manager (again, more is worse, and there is a big jump from the condition where the maximum is 1 anywhere in the chain and the condition where the maximum is > 1.) [1] this includes both “people” and “functional” managers where those roles are distinguished, as well as people that aren't characterized as having reporting relationships but from whom the employee is expected in practice to take direct direction.
ReptileMan 7y ago
>It's simply not possible to have 100,000+ employees and less than 6 levels, unless you have managers with tons of employees But I think that MS had 13 levels (IIRC joel) in 2006-ish, while the company had 7 in 1994 when he arrived ...
jvanderbot 7y ago
What I instinctively interpreted that as is "branching factor". i.e. the number of subjects at each management node. Tall companies have many middle managers (middlemen) but small teams with no slack. Flat companies have large groups and infrequent interaction with leads or managers. Perhaps more slack, perhaps more ambiguity and less direction. Both can be unhealthy.
IfOnlyYouKnew 7y ago
There is also a “toxic valley”, where an org has outgrown what can be managed by informal social mechanisms, but has not (yet) had the time/reached the size for professional practices.
- gadders 7y ago
  I always liked this essay about "The Null Process" [1] which touches on some of the same topics: >> Our fear of "unnecessary process" has created workplaces with something worse than bad process: no process. I call these types of processes "null processes," and they are rampant at startups and technology companies. [1] https://www.kateheddleston.com/blog/the-null-process https://www.kateheddleston.com/blog/the-null-process

awinter-py 7y ago

would be interesting to see some AI-assisted version of holacracy replace middle managers -- you'd still have low-level team leads and high-level strategy people, but a lot of the informational glue and project updates could be taken over by machines you'd need to go into a project like that (1) believing that middle managers eventually suppress more value than they add and (2) you're okay operating in a random mix of cowboy team leads and inconsistent organizational glue between them. you'd set broad parameters like 'don't over-duplicate work', 'pick projects that are more or less on-topic with our company', 'get third party verification of plans and estimates'. Firing could be some mix of 'future needs', performance and random. Moving teams would happen based on an internal market of skills & personnel budget. might be chaos but it could be fun.

Matticus_Rex 7y ago
I've been really interested to watch what Holacracy actually does (sans AI) in healthy, agile companies. A lot of the management overhead gets cut (while keeping functional management of the company, though a lot of it becomes process rather than people), and a lot of the projected issues are just things someone can propose a solution for and get it enacted via Holacracy governance. Springest is a really interesting example. Very anti-bullshit culture, rigid focus on OKRs and efficiency, and yet they're extremely light-hearted and lack the political crap most similarly high-performing companies have. Maybe AI makes it better in the long run, but I'm a believer already.

Misdicorl 7y ago

> It's simply not possible to have 100,000+ employees and less than 6 levels, unless you have managers with tons of employees Math does not check out. 5 levels for 100k employees results in 10 reports per manager (assuming a top level with 10 people rather than 1, which is approximately correct given CEO/CTO/CIO/etc/etc/etc). Maybe 10 qualifies as 'tons' for you? Going down to 4 levels gives 18 reports per manager, which feels a little closer to 'tons'.

dvirsky 7y ago
10 direct reports on average is a lot. It might be okay at the highest levels where you mostly make strategic decisions, but it's too much for a manager that's involved in day to day processes and decisions.
chadash 7y ago
I think we agree in principle, just different understanding of how to count levels. Based on the article, a company with a boss and one employee is two levels ("With only two levels, a boss and those who report to the boss..."). So for 100,000 people, with average 10 employees per manager: 1 Big Boss (Level 1) 10 (Level 2) 100 (Level 3) 1000 (Level 4) 10000 (Level 5) 100000 (Level 6) That's six levels. In practice, some people will have more than 10 and some less, but unless the people under you are doing menial work, I think 10 is probably middle of the road.
- Misdicorl 7y ago
  Fair enough if you want to include the one person level. It seems rare that any large company uses a structure like that though
- Aperocky 7y ago
  The math stop checking out at level 4. You have level 5 managers managing 10 each, and you have level 4 managers managing 10 teams each? That stops checking out. The higher you go the more disciplined you have to be about grouping relating teams under the same umbrella for efficient execution. In fact, the levels don't matter as much as the autonomy at each level. The more autonomy, provided that the right people are hired and are focused on the goal, makes it better and less managerial burden going up the chain.

throwawaypa123 7y ago

the 6 layers isn't really true, in part due to line managers can manage 15-30 people. If everyone is doing the same job that is fairly rote a single line manager can handle a lot of people, reducing layers. Additional CEOs tend to have a lot of direct reports. Sample Example (0 Layer) CEO --> 10 Direct reports is typical for large publicly traded companies. (1 Layer) C-Level --> 6-8 (10,000 / 8 --> 1,250) (2 Layer) SVP / VP -- > 6-8 (1250 / 6 --> ~200) (3 Layer) Middle management (200/ 6 --> ~30) (4 Layer) Line managers (1-25)

dragonwriter 7y ago
> the 6 layers isn't really true, in part due to line managers can manage 15-30 people Yes it is, IMO; any layer above line managers over nonprofessional staff should have a span of control not greater than about 3-5 subordinate managers with full-depth (1 level less than the manager) organizations and a similar number of supporting staff that either have no reports of their own or supervise organizations about 2-3 levels less deep than the manager above them. Yes, lots of real organizations have broader spans of control, but lots of real organizations also are what the author describes as “immoral mazes”.
dtnewman 7y ago
From the article: > With only one level, there’s nothing to worry about. With only two levels, a boss and those who report to the boss, the boss has skin in the game, no boss causing problems for them, and not enough reason to reward bad outcomes. With three levels, there are middle managers in the second layer, so one should be wary. Based on this wording, it seems like the CEO and the people under the line managers count as levels (levels though maybe not levels of middle management). By that reasoning, this structure would indeed be 6 layers. In reality, it varies by division if you are in a large company. If you join at the bottom of the legal department at Google, you probably have fewer layers above you than if you are a junior dev working on android.
spoondan 7y ago
There may be situations where 15-30 direct reports is workable, but that is a very large number of reports. Academic studies generally settle on a maximum effective team size around seven, give or take a few[1]. In my own experience, I saw diminishing returns and reduced satisfaction in my management performance after ten direct reports. These issues resolved with a restructuring. But you don't have to take academic studies or anecdotes to heart. Some back-of-the-envelope calculations suggest there are severe challenges to effectively and healthfully directly manage 30 people. Take the typical 40 hour week[2]. With 30 direct reports, that gives a maximum of 40/30 = 80 minutes/week to devote to each individual employee. If the manager does a weekly or bi-weekly 30 one-on-one, that leaves about an hour on average to deal with any issues. Of course, not every employee has high priorities issues at the same time. But we've only considered the individual, one-on-one work of the manager. There is also the team-wide and cross-team work. And, it turns out, team size also dramatically increases the time and effort these other facets take. Consider the typical daily stand-up (blech), used to keep the team aligned. If the manager allots (and strictly enforces) five minute slots per team member, that meeting still takes 2.5 hours. People won't be happy with that, and it will cut into the managers limited time to actually make progress on individual and team issues. Even if the manager splits the meeting (to spare team members for a super long meeting), the manager needs to attend each one to stay abreast of the big picture. Likewise, we presume that if a team has 30 reports, it's because what they are working on is of critical business importance. There are more demands from sibling teams in the same org, more budget and staffer reviews with higher-ups, more hours spent with HR and recruiting. (Alternatively, the manager is empire-building. But that doesn't really help the calculation. The time that the manager would've spent on delivering business value is instead spent on politics.) Even if you think you can keep all these balls in the air, it's hard to not conclude there are incredibly tight windows here. And the whole thing is very fragile. If you fail to identify, understand, and address a critical issue, it can quickly infect other team members. [1] See, for example, https://knowledge.wharton.upenn.edu/article/is-your-team-too-big-too-small-whats-the-right-number-2/ https://knowledge.wharton.upenn.edu/article/is-your-team-too... [2] I know. We often work more than that. The exact number doesn't really change the story, especially since we're talking about retaining our physical and mental health.

alecbenzer 7y ago

Given the author admits: > Note that those outside the company, such as investors or regulators, seem like they should effectively count as a level under some circumstances, but not under others. I think there's obviously some room to fudge with what the exact boundaries of the "organization" are. So yeah, maybe a sufficiently-empowered department head functions more like a CEO/boss.

dragonwriter 7y ago

> Number of levels in the hierarchy seems like a bad way to determine this. It's incomplete as a measure of the problem, but it's definitely a valid measure of a key risk factor. The span of control (# of direct reports) of each supervisor in a direct chain of command from the employee to the top-level manager is also a measure (higher is worse), as is the “span of reporting” (the number of supervisors[1] to which each employee reports) from the employee up to the top-level manager (again, more is worse, and there is a big jump from the condition where the maximum is 1 anywhere in the chain and the condition where the maximum is > 1.) [1] this includes both “people” and “functional” managers where those roles are distinguished, as well as people that aren't characterized as having reporting relationships but from whom the employee is expected in practice to take direct direction.

ReptileMan 7y ago

>It's simply not possible to have 100,000+ employees and less than 6 levels, unless you have managers with tons of employees But I think that MS had 13 levels (IIRC joel) in 2006-ish, while the company had 7 in 1994 when he arrived ...

jvanderbot 7y ago

What I instinctively interpreted that as is "branching factor". i.e. the number of subjects at each management node. Tall companies have many middle managers (middlemen) but small teams with no slack. Flat companies have large groups and infrequent interaction with leads or managers. Perhaps more slack, perhaps more ambiguity and less direction. Both can be unhealthy.

IfOnlyYouKnew 7y ago

There is also a “toxic valley”, where an org has outgrown what can be managed by informal social mechanisms, but has not (yet) had the time/reached the size for professional practices.

gadders 7y ago
I always liked this essay about "The Null Process" [1] which touches on some of the same topics: >> Our fear of "unnecessary process" has created workplaces with something worse than bad process: no process. I call these types of processes "null processes," and they are rampant at startups and technology companies. [1] https://www.kateheddleston.com/blog/the-null-process https://www.kateheddleston.com/blog/the-null-process