This is an overview of my review ratings and in general my review philosophy. Now, on the one hand, all products are relative... even a game which is poorly done on almost all fronts may have its adherents who claim it is a fun game. I try to respect that different games are trying to do different things, but I still have an objective rating for how well the game achieves what I believe it is trying to do.
Most reviews have a rating system, and readers will expect some sort of rating. Thus, while I don't think much of ratings in general, it is standard enough that I will include very rough ratings from 1 to 5 stars. If you have any interest in the product, the full review takes precedence over the simple rating.
In genreal, one star indicates that the product is nigh unusable -- of interest only if you are a collector or have some very special interest. Three stars indicates that the product is roughly even with the industry standards, such as the average GURPS product or White Wolf Storyteller system product. Five stars is the hypothetical perfect -- a game that is worth buying just to read it, even if you have no interest in the genre or topic.
This is a rating for how well the product displays its contents. For a rulebook, important features are readability and organization -- especially how good the table of contents and index are. For a setting/genre book (which may be the same product), the art and flavor text are also important. Art is judged not just on how good it is, but on how well it illustrates the game-world in question.
Of course, it is not possible to entirely separate presentation from content. However, I try to make a stab at it.
This is a rating for how well the game works in play. Even moreso than Product / Presentation, this rating is a matter of taste. As with the other, this rating applies to both the rules and the setting.
The rules are judged in terms of how well they work relative to the desired level of detail. i.e. A detailed combat game might have complex modifiers for a given rifle shot. The question is, if you want that level of detail, how well do the rules work?