When I am on a conference program committee or editor of a journal, here is how I choose a sub-reviewer who I ask to read and evaluate the submission.
First, I carefully read the part of the introduction that talks about "related work" and about "techniques", looking up every reference. Then, I quickly look through the body of the paper, checking whether some paper is referenced inside the proofs. Then I know that that paper is probably important and the authors of that papers would probably have a well-informed opinion on the submission.
Once I have my list of relevant or very relevant references, I restrict attention to recent papers, discarding anything older than 10 years old, because I assume that, absent other information, the authors of those papers are probably no longer interested in the subject. I look at redundancy, giving higher priority to authors who have written several of the relevant cited papers. I look at conflicts of interest, giving lower priority to people who have recently co-authored papers with the authors of the submission. Then, I give higher priority to authors of papers that have been published in similar venues, because I assume that they will be better able to evaluate the quality threshold. Finally, I try to pick a mix of junior people, who have more time and may be more interested in the submission, and senior people, who may have more perspective and more of a long view on the results submitted.