I think most firms that tried it have since scrapped it. It turns out to be about as effective as firing a random 10% and promoting another random 10% of staff.

Consider: a project is floundering. You put your best engineers on it to try and save it. It fails anyway. They get fired! Yay stack ranking.