I find that diagramming, while it can be useful, is very time-consuming and, frankly, not entirely necessary. MGMAT's approach is particularly cumbersome and seems to imply that we should imagine all possible weakeners to a question, which is not efficient; Kaplan also encourages us to rewrite the Conclusion, etc., which can be useful for some non-native speakers, but I generally find it's a bad plan to rewrite what's already on the screen in front of you.
One thing to keep in mind is that, on the LSAT, students have to do about 28 Logical Reasoning questions, which are generally harder than GMAT CR questions, in only 35 minutes. This requires much more efficiency in thinking through such questions, and unless there's formal logic involved (there never is on the GMAT), writing anything down is discouraged. Well, having worked with a lot of LSAT students and a lot of GMAT students, I'm pretty sure that the LSATers aren't smarter; they just get trained to do it without writing anything down. As a result, I've started to encourage GMAT students to also not write anything down. You can keep the salient info in your head; you just have to be actively asking yourself, as you read, what info is salient and what isn't. This takes practice, but it's doable.
Another thing that really helps is anticipating the answers you'll see, both right and wrong. My biggest advantage in CR is that I've been doing this for 5+ years and any new CR question I see, I know I've seen something like it before, so I've learned to expect certain kinds of right answers and can quickly eliminate wrong ones. This kind of confidence, too, takes practice, but it makes things far faster if you know as soon as you finish reading the Stimulus what kind of right answer you'll see, and know as soon as you read some out-of-scope word that a choice can be eliminated.
