Eventually it became clear that performance here was based more on perception than output. How loud and visible you were in meetings mattered more than what you actually shipped. You could ship a major feature being promoted all over their social media and be told you weren't doing enough because you weren't being loud and political about it.
Then came WorkWeave.ai, an AI tool that scores your code output and estimates how long an expert engineer would take to do the work. Nobody really understood how it worked or how to move the score. Leadership started using it as a performance monitoring tool despite it literally displaying a banner saying the data was incomplete. PIPs started going out based on these scores.
I was one of the lucky recipients. A month later, the CTO stated in an engineering-wide meeting, with complete composure, that he didn't care about WorkWeave scores and they weren't being used to evaluate performance. I had a PIP document in front of me with my WorkWeave score typed on it and the exact number I needed to hit to get off it. I've seen more convincing performances, but not many.
Multiple engineers started quietly looking for exits as a result, and a few beat me out the door.
It later came to light that the PIP wave was actually a compromise. The original plan was to lay off more engineers than those who ultimately received PIPs and may have included brand new hires.