In the world of the rhetorical triangle, my audience is an ill-thought-out metaphor.
So, I'm perusing Amazon. There are two items I come across, and one's ratings are significantly higher than the other. I have no other prior information on either book, and something in Amazon's algorithms directed me towards both, so I'm not even entirely sure what they're about. I read the descriptions, and they strike me as fairly similar. Both descriptions could be of a bad, textbook romance, or of a well-written tale of what "romance and intrigue" would bring to mind if it weren't so thoroughly associated with textbook romances #8-47, 89, 307-415, and 112. I pick the one with the lower rating.
Book set #2 looks almost exactly the same. If there are any differences in rating, they are in the hundredths. I was directed in the same way, and one looks textbook romance, one looks different in an interesting way. I pick the one with the higher rating.
The point of those paragraphs was to say that the average star rating doesn't factor in. Honestly, the stuff I like is probably going to do terribly in my age bracket, and the ratings say more about the advertising than the actual books--if you get a ton of adolescent males suddenly getting given Twilight books, guess whose rating drops? If the book is for an incredibly narrow audience, and no one else will understand it, but it also happens that only three people--myself included--outside that group will ever find it, then its rating will probably be pretty decent. And polarizing books have 2.5, because half the people give 5 and the others 1 (or 0). It just doesn't work for me.
I start out looking at low-rating comments, seeing if I care about them. Then I look at high-rated, same metric. Basically:
1. Spelling. Not just proper spelling--it's the internet, people make mistakes and don't edit them. Big whoop. But if I come across a review that's written in 1337 or with a bunch of random symbols in the middle...I'm not part of that demographic. It's probably not going to say anything I care about.
2. Grammar. Once again, grammatical errors happen. The ones I'm looking for here are the ones where they wear their poor grammar as a badge of pride--"I know my grammars gonna be bad, but u can just deal."--and false intelligence. The most common are putting "I" where "me" belongs, or "whom" for "who". "He got the book for me."-->"He got the book for John and me." And if the who is performing an action in active voice, that's a who. "He is there."-->"Who is there?" "Whom is there?" just tells me you're trying to sound smart and failing.
3. What are you saying? All of these have been subjective, but this is probably the most so. Say the person says that the author went too far into the background and mythology of the piece, to the point of most of the story taking place in the past. I find that fascinating. If it goes dry, "Oh, by the way, vampires can move at exactly thirty two feet per second per second horizontally, and werewolves can move a 8.6753 meters per minute while in shifted form, but..." then I'm probably not interested. But if someone is complaining about background as a bad thing in and of itself, we have different tastes. That review doesn't help me, so I move one. Positive reviews have the same thing. If the person talks about how the graphics in X game are so amazing and immersive and...sorry. Graphics, not really my bit in videogames. I want to see everything clearly, so crisp graphics are good, but I only really fall into the sun-dappled autumn leaves fluttering if I'm not worrying about winning/losing/playing.