The numbers are discussed by upper engineering management. They might change over time depending on the situation the company is in.
Think of a hypothetical organization with 10,000 software engineers. You're the head of engineering. You have ten divisions, each with a VP managing 1000 engineers. Each VP has ten directors, each managing 100 engineers, and each line manager has 10 engineers on the team.
The engineering team reviews the past year. You notice, overall we fired 500 people last year. Okay, so on average that was 5% of staff. Seems reasonable just as a sanity check. (You ask your buddies at other large tech companies, and the other heads of engineering are reporting similar numbers.)
Now you look through the individual teams. A lot of 10-person teams don't fire anyone. That makes sense. But would you expect a director to fire nobody from their org? From 100 people... well, maybe. I'd be a little suspicious. I'd ask some other directors, does this person have a reputation of a very high quality team, or is it more likely that this director is lax, and their org doesn't manage out its underperformers?
Now imagine a VP fired nobody. 1000 people and they all were high performers. Yeah, that doesn't seem right. That VP is probably letting their team get away with low standards. If you were the head of engineering meeting with your VPs, I think the group would be able to come to a consensus of, there's a problem here. It's based on the 5% target but it's not a hard and fast rule.
In the long run, having a high-performing team is better for morale than firing nobody. It's the difference between working at Meta and working at the DMV.
So: You have no measure of performance and go off vibes. You have no scientific basis and go off vibes. You pretend this is all logical and common sense but it's really all just vibes.
You could end up performing way better firing half of your teams. And what about your managers? How is an underperforming manager in the higher positions ever fired?
I get that measuring performance is incredibly difficult, but dressing it up as if 5-10% of people are underperformers anyway is just so tantamount to how baseless and incompetent most businesses are run these days. Especially in software.
Think of a hypothetical organization with 10,000 software engineers. You're the head of engineering. You have ten divisions, each with a VP managing 1000 engineers. Each VP has ten directors, each managing 100 engineers, and each line manager has 10 engineers on the team.
The engineering team reviews the past year. You notice, overall we fired 500 people last year. Okay, so on average that was 5% of staff. Seems reasonable just as a sanity check. (You ask your buddies at other large tech companies, and the other heads of engineering are reporting similar numbers.)
Now you look through the individual teams. A lot of 10-person teams don't fire anyone. That makes sense. But would you expect a director to fire nobody from their org? From 100 people... well, maybe. I'd be a little suspicious. I'd ask some other directors, does this person have a reputation of a very high quality team, or is it more likely that this director is lax, and their org doesn't manage out its underperformers?
Now imagine a VP fired nobody. 1000 people and they all were high performers. Yeah, that doesn't seem right. That VP is probably letting their team get away with low standards. If you were the head of engineering meeting with your VPs, I think the group would be able to come to a consensus of, there's a problem here. It's based on the 5% target but it's not a hard and fast rule.
In the long run, having a high-performing team is better for morale than firing nobody. It's the difference between working at Meta and working at the DMV.