like literally any social situation involving applying for something and being admitted, connections and networking is so important. i used to think one way going to grad school is different from getting a job in finance, etc., is that academic success is highly valued over networking.. however, ive come to understand the reality of this world we live in, and you can't discount connections and networking.
go to a good school, work with renowned professors that have widespread connections, get an amazing rec letter, and youre good to go. of course, granted that you fulfill some standard in academics (high gpa, good gre scores, etc. but i feel like almost everyone who applies for physics phds have that under control).
the thing i learned about the pgre is (after taking it twice), the score is definitely not an insignificant factor in applying. especially if youre going for the top tier programs, pgre is definitely very important, and as someone stated earlier, i would deifnitely recommend using the time to study for pgre that youd use for gre.
but in the end, i think the most important thing i observed from this years admissions cycle over the last (i applied last year also, but failed miserably) is the importance of how well your research interests/previous experiences will match the school/professor youre applying to. i learned the hard way that you cant just randomly decide to switch to another track within physics if all your research experiences and abilities come from a different track.
so my advice to future applicants would be to get three things down (again, i assume that people who apply generally have high gpas and other academic achievements): networking/connections, pgre, and applying to the correct research interest/professor program. in the end, the profs are paying you to do research, and theyre not about to splurge their fundings on some rando.