First priority is finding flaws. These are mostly objective. There are the obvious ones within the diff like missed edge cases, corner cases and unexpected behavior changes. Then there are less obvious ones, in which you have to think about the change in the context of the greater system. For example, checking for things like side effects, backwards compatibility and rollback safety.

Second priority is finding readability gaps. These are mostly subjective. Things like naming, structure, bad abstractions, over-complexity and intent. These are important, but providing feedback on them is a bit more nuanced. This is where the human element of code review comes into play.