As a note people with 100+ direct reports are not really managing them. Often it'll be indirect as in there are 100+ people in a hierarchy below you. You might only mange 10 people but they manage 10 people and so on.
In terms of reporting your "team" is all 100 people even though you may have never interacted with half of them other than an introduction.
FYI twitter seems to be moving to a low manager high employee count. Musk himself said that a ratio of 1 manager per 10 coders is way too high. I suspect he wants it at 10x that amounts.
My current manager has 31 direct reports.
Sorry if this is confusing it's a hard subject to describe over text and I think there is a lot of nuisance lost over text here.
should/do they consider other metrics: revenue, active users, etc under management?
On one hand, those are astronomically high revenues. Great! On the other hand, I'm losing $100M per year. Suck! But I was brought in to fix things up after some idiot who ran things into the ground. I'm doing great! But it's a growth market; maybe it's because of that? Suck! But in fact, I'm bleeding money for growth. Great!
... and so on.
So all those other things can be spun. It's nearly impossible to objectively evaluate executive performance.
They definitely show up on OKRs and similar, which can be managed by setting low objectives.
Most problems are better-solved by small teams, but that's usually not how incentives align.
Above some level in the corporate ladder, executives understand these games and play them completely cynically. It's easy to become a manager without this. You don't get to be in the C-suite at 10,000 person firm without playing these games near-optimally.
Note that this is not the only part of the corporate ladder game. Other parts may keep this (somewhat) in check, so you usually don't have completely pointless 5,000 person divisions your local supermarket branch.
They do less well for keeping this in check at monopoly-profit firms like Meta. In monopoly-driven firms, it's really easy to start politically-popular pointless units (I suspect, in this case, a skunkswork, forward-thinking division engaged in something with no real corporate value, so long as it aligns well with a buzzwordy-topic like AI/DEI/VR/etc.).