Lots of interesting information here:
>The agentic era affords GitLab the largest opportunity in our history as a company, and we're making the structural and strategic decisions to meet it
>Operationally, we grew into a shape that was right for the last era and isn't right for this one
To meet their largest opportunity ever, they believe they need less resources. I'm not sure I understand how that follows.
>We're rewiring internal processes with AI agents, automating the reviews, approvals, and handoffs to speed us up
Is this also in the list of "we create code twice as fast and the bottleneck is review so YOLO no bottleneck?". I've yet to see a convincing justification for this. If anything, if you're going full throttle all the more reason to watch the steering wheel, no?
That said, 8 layers of management is a lot of management, and every line of the message seems like leadership truly believes they are sinking in bureaucracy. Let's see how unneeded those 3 layers they're cutting were.
> 8 layers of management is a lot of management
Seems like a fair assessment. Maybe they should start by getting rid of the people who put that structure in place?
Didn't they do that? Staples only came in as CEO at the end of 2024, and I assume he has been working on a plan to restructure the company since then. Because their financials are not great, and they have been losing money every year since 2019.
I don't know about gitlab, but tech companies (Meta and Grab) tend to hack off the bottom of the management chain, instead of cutting off the top (aka as the people that created the 8 layer system).
bottom level teams are merged to form larger teams.
Yeah, they never fire the VPs and SVPs in this process. Just a bunch of the hard-working line managers who are actually involved in the day-to-day engineering work
Companies are shaped more or less like pyramids. If you want to cut a meaningful amount of people from the organization, there's just not enough of them at the top.
If one person at the top of the pyramid earns 100 times what people at the bottom earn, cutting a few of them is still meaningful. Also, cutting a single/few person(s) that are mismanaging the whole organization is extraordinarily valuable too.
But salaries - sorry, operational costs - are an inverted pyramid.
You know how many layers of management I have Bob?
You have to translate from management speak: They see the largest opportunity ever to make more money, by spending less money on staff.
8 layers of management???
At gitlabs team size, that means every manager has 2-3 reports? Yeah, I'd be cutting layers too.
https://handbook.gitlab.com/handbook/company/structure/
> GitLab has at most eight layers in the company structure (Associate/Intermediate/Senior, Manager/Staff, Senior Manager/Principal, Director/Distinguished, Senior Director, VP/Fellow, Executives, Board).
> [...] You can skip layers but you generally never have someone reporting to the same layer (Example of a VP reporting to a VP).
So they're counting the board of directors as a layer above the CEO.
I'm speculating, but they probably also have an unbalanced tree - you'll often see the IT security chief reporting directly to the CEO (because it's important to keep on top of, and they need authority to do their job) but only having 50 people below them in the org chart.
In some corporations you also sometimes get almost-nonexistent ranks created to smooth over a reorganisation. If a level 5 bureaucrat decides to merge the departments of two of their level 4 bureaucrats, they could demote one of them. Or they could make one into a level 4.5 bureaucrat.
this is the typical layer at a big tech co.
you can simply cut out so many layers of management & coordination tax & remain with a leaner organization.
That’s not what layers refers to. What they mean is how many managers between the CEO and an employee. Made up Example: CEO->CTO->VP of Infrastructure—> Director of Platform-> Sr Manager of AWS platform —> infra engineer would be 6 layers
Gitlab has approx. 2300 employees (!)
I never really got why they need to be a public company in the first place.
That's mad. Gitlab feels like software made by 50 people. Consider how unpolished it is and how slow it is moving.
You're onto something, I've been part of enterprise projects with a <10 engineer team that had more polished and diverse feature offering.
Did you typo an extra 0?!
They have 2800 employees with 5-10 reports per manager.
At 8 layers of management (so 9 layers total, with the bottom rung being non-management), 3 reports per manager comes out to 6561 employers on the bottom rung. At 5 reports each, that 8 layers would give you over 300k at the bottom, an 10 each would give you 100m at the bottom.
> We’re flattening our organization because eight layers is too deep
Eight layers total
So... That's the assumption that would give you 300k employees with 5 reports per manager. Sum of 5^x for x in [0, 7] is ~300k.
The GP miscalculated it.
Mathematically that would work out to a lot less than 8 layers of management.
I wonder if they have 5-10 employees per manager at the bottom of the org chart, but a lot of middle managers and manager-like titles mixed through the middle.
If its anything like the other tech companies, you'll have a bunch of overworked low-level managers with 20+ reports each, and then somewhere up the chain you'll find directors and VPs chilling with 1-2 reports
Solution: fire half the line managers, and make the rest also do IC work.
If anyone has a VP-level position open, I'm willing to send you my resume. There is a salary level at which I am willing to do work entirely without shame.
That's a really crazy number of employees considering they have one product that barely seems to change and is at best on par with similar products created by comparatively miniscule teams (Phabricator, Forgejo).
I'm on board with your gut that this feels more YOLO than careful but to be fair, in the engineering world fly by wire is very much precedented. I'm specifically thinking of the B2 bomber where it's essentially unflyable without a computer between the inputs and the outputs. Partially just keeping the plane from turning into a frisbee by reacting faster than a human possibly could, but also treating the controls inputs as the intent and manipulating the control surfaces programmatically in order to make that work. It's not quite the same thing of course but I think there's some carryover.
Still. Not a huge fan of this announcement or the general ways the landscape is evolving these days.