CompSci academic thought tech support was useless – until he needed it

I feel like this argument only holds weakly. By providing "numbers" you think you are providing objective points, but how to weigh incidents solved vs service vs other metrics is still purely subjective.

Ultimately what you are doing is providing incentives for certain actions. The problem is that as soon as your metrics are established they can be gamed or misdirected

To be fair, it's not often a person screws up consistently, never catches the error, never contacts anyone through an alternate route to inquire, builds up months or years of ire, and then makes a public spectacle of themself.