It's for validating data, in general. A huge percentage of interoperability issues between clients relates to missing tags or missing data within tags.

Validation and QA are different things tho right? Or am I thinking about this incorrectly

Not sure, I was just primarily responding to @Arjen since he linked to the nostrability repo that hasn't been updated in almost 6 months. Wanted to clarify. If Client A puiblishes a NIP-22 events without a root event tag, then Client B (and perhaps event Client A) won't be able to find it. Schemas can catch this error either in production or in a CI/CD pipeline. It can also detect these issues by fuzz testing events. I would personally consider that QA, but maybe not?