All these big companies give grade employees on a curve. So, in Meta parlance, only 5-10% of employees can be "Strongly Exceeds Expectations". This is measured across the director level, so units of 100-200 employees.
Here's the problem: at Meta 10-15% of employees in these buckets has to get a subpar rating. This includes Meets Most Expectations (2 MMs in a row probably means you're on a PIP), Meets Some (probably auto PIP), Meets None (fired) and Non Regrettable Loss. This last one applies to rating employees who leave.
NRL in this bucket may have changed. For one thing there was talk that HR not management was responsible for NRL designations because otherwise orgs made up their quota of subpar ratings by using leaving employees as ablative shielding.
Don't believe me about these quotas? Two words: "tbgs genTargetPercentage".
I personally don't have issue with avoiding ratings inflation in principle. While I was Google I don't believe subpar ratings had a quota like FB/Meta has (or at least had).
This is classic "up or out" GE nonsense. And I can tell you from experience that this is incredibly toxic. The entire calibration process devolves into a popularity contest and rewards those who are good at politics. While in a growth period, these problems didn't tend to matter.
I said it when the layoffs were announced: those who were laid off were largely the lucky ones. More layoffs would be coming and those would be with forced attrition. This is what this is. Subpar ratings hurt your bonus and equity refreshes and and will make it more difficult to transfer teams. All of this is by design. You're in that (now increasingly wider) band of employees they want to quit rather than fire.
Disclaimer: ex-Facebooker.