I read Andrew Wheeler's post on this topic a few days ago and, as I write book and movie reviews, thought it was an interesting topic. I also want to point out the New York Review of Science Fiction's Guidelines, that someone recommended I read back when I started reviewing books for SF Signal.
First off, what's the point of a review? To give enough information so that readers know if the book will be interesting to them. You can use many methods to achieve this: stars, pro/con sections, comparisons with other books, etc.
I personally find some of these more useful than others. I don't generally rate things with stars because I've noticed different sites give different star ratings different meanings. And it's easy to assume a book is better/worse than it is based on the rating. Here's a comparison chart based on several popular sites:
friend me on GoodReads and see what I'm reading now, rather than learning it when I post the reviews several weeks or months later, feel free.) I'd also argue that there's a difference between saying a book was bad and saying I personally didn't like it. Some books can be well written with lots of good elements and simply not appeal to me as a reader. I'd hate for people to read one of my reviews and dismiss a book they might like because I rated it a 3 or below. Everyone has different tastes, and ratings don't give you the information you need to decide why a book 'wasn't good' or if YOU would like it.
That's why I borrowed SF Signal's pro/con idea. Not only is it a good way to get my thoughts in order when reviewing, I feel that it's helpful to know what elements the reviewer did or didn't like. If you like slow reads and I don't, well, my con is your pro - you can use that information when reading the review proper to see if other aspects of my taste are similar to yours. If they're not and I gave the book a negative review, maybe it's a book you'd like.
As for comparison books, I'm on the fence with its helpfulness. If you've read the comparison book(s) then it's helpful, if you haven't... And what aspects of the books are similar? When I do my New Author Spotlight posts I make 3 comparisons. I generally have to pick one element (character, subgenre) and find other books that match that, even if they don't match in other ways (like tone, pacing, etc.). For example, when doing Will McIntosh I picked other books that had a soft apocalyptic feel to them (or at least, books that showed the world in the midst of a collapse - of which there weren't many to choose from, most authors preferring to do post apocalyptic or war style apocalyptic books). When it comes to comparisons there's an element of guesswork, as even if the themes and genre are similar, you may love one book and hate the other.
Ultimately, if reviews I write help you decide if a book is for you, then I'm doing my job correctly.
What aspects of reviews do you find helpful? How do you decide if a reviewed book appeals to you?