A crawl can produce thousands of warnings. That does not make all of them important. I start with the commercial pages, the search demand they serve and the behaviour that prompted the review.
If organic landing pages disappeared after a release, I compare what changed. If a new store is creating millions of filter URLs, I trace how they are generated and linked. If Google sees an incomplete JavaScript page, I compare the server response, rendered HTML and browser behaviour.
The aim is to separate evidence from tool noise and give the team a sensible order of work.