Genette defines and explores the paratexts of a text. Under his definition, paratext is essentially a threshold, the parts of a book beyond the explicit content that help the author convey his meaning. That includes elements included in the book (the peritext) and those outside (the epitext). Peritext includes the title, notes, prefaces, and so forth; epitext includes journals, correspondences, press releases, reviews, and so forth. And while a generation or two postmodernism has frowned at the notion of authorial intention, Genette defends his inclusion of authorial intention in the definition of paratext, on the grounds that authorial intention still has to be squared with, even if it's not the the only focus, and that the limits of paratext must be drawn somewhere. That seems to be a key point to me; paratext is a phrase that's been taken up in other studies, including film, television, and video games, and further authors tend to expand the term significantly. Genette recognizes that if it's to have any meaning, it must have limits, and while I disagree with the limits he chose, I agree with the general intent.
The book itself is somewhat scattered, though it's largely scattered by design. As such, I can't really advise a thorough, chapter-by-chapter reading, unless you are very, very interested in how Balzac used subtitles, for example. To get the general thrust of Genette's argument, I would instead recommend the introduction, the conclusion, and a section on the chapter on notes, where he states that the operating question for a scholar should never be whether a paratext exists, but whether it's useful to consider the text in that context.