Ha! (to your last part.) So sorry to have assumed! (and I totally didn't look at which forum this was, so no virtue on my part...). I completely retract.
Not to get all meta on you, but I think some of what you call inefficiency is inevitable revision. No one gets writing right the first time; you always have to go back and rework. You're not going to be able to avoid realizing, hey, this issue is more important, I need to go back and add it. And once you add it, you'll probably have to re-do some of the organization. That's inevitable. And personally I do a lot of rereading and rehashing of cases - I never get it all on the first read-through, and I don't think anyone can. However, I have always been of the "spew material on the paper, then clean it up" school of writing; pre-planning doesn't work very well for me, and revision is crucially, crucially important.
When I was in LS, I found Scrivener a really helpful piece of software, because of how it allows you to organize your materials; I could import all my cases and organize them by issue, and draft different sections of the piece separately and move them around before committing to a particular organization. I don't use it now mostly because it's not supported anywhere I've worked (and I didn't think it would fly to save materials in a format no one else uses). But I miss it.
This is what I usually do now: I just start with a .doc file of plain old notes. I create basic headings (if it were for the Lemon test, they would probably be "secular purpose," "advance/inhibit," "entanglement") just by typing them in bold. Under each one, I list the cases that are relevant with either a paraphrased parenthetical (say, "prayer at football game fails secular purpose prong") or a quote from the case (I often cut and paste big chunks so I have pertinent language at my fingertips). If the topic's really complex I sometimes separate cases according to in-jurisidiction, out of jurisdiction-state, out of jurisdiction-federal. I just look for cases and read them to begin with (or treatises or statutes or whatever).
Once I start to feel like I've really synthesized that raw material, I usually start writing rule statements with cites/support. ("The secular purpose prong under Lemon requires...." See Smith v. Doe, 12 F.17d 63 etc.) And then I'll apply them. ("Here, the secular purpose prong is not met because...") I don't do this in any particular order; just as the understanding comes to me. So I might write up a whole section about the secular purpose prong and how it applies to the facts at hand while I'm still reading/researching about entanglement. (When to stop researching: when you keep coming across the same cases/rules.)
Once I've come up with all these kinds of building blocks, I compile them in whatever way makes the most sense. So, for instance, the Lemon test is usually written with the entanglement prong last, but if the entanglement and advance/inhibit analyses are really easy in this case, I might put them first just to get them out of the way (e.g., "Here, the entanglement prong is satisfied because... Likewise, the advance/inhibit prong is satisfied because... However, the secular purpose prong is more complicated [and go into a more detailed, extensive discussion of secular purpose].") Also, the order I put them in the writing has nothing to do with the order in which I wrote them.
I frequently think I will need more or fewer headings/subheadings than I end up using. I might think, oh, three headings for the three Lemon test prongs, but then when it comes to one of them, there may be multiple tests for a particular prong and so I end up needing to subdivide further and have different subheadings for each test (my analogy's breaking down because I don't remember much about Lemon!). I don't usually make those decisions ahead of time, I wait till I've done most of the research and am writing. But some of that depends on how well I know the area; some issues that I know well, I know ahead of time what I need to cover, and it's easier to pre-outline than if I'm dealing with a new area of law.
Personally, I would fail miserably at writing out the first sentences to each paragraph before writing, because my brain just doesn't work that way. I don't outline in that much detail, because I can't; I don't figure out what I'm going to say ahead of time, I have to try to write it out before I know what I'm arguing. I do sometimes roughly outline sub-issues, but it's usually because I've been trying to write about something and it's not working, so it's more taking the material I already have and organizing it, rather than picking an organization before I have the material. But other people work differently. While I think people can learn and adapt various methods, at a certain level I think you have to accept the way your mind works. Even from grade school, I never outlined before writing (and in fact, I didn't really outline for final exams in LS). But I do know some people (not in law) who have to work everything out in their head before they put it on paper, and once they do, they do very little revision on paper. That sounds more efficient, but it usually doesn't take them any less time than it takes me, overall - I just start putting things on paper sooner.
I have no idea if that's at all helpful. And sorry to go on so long - I just have spent a lot of my professional (pre-LS) life writing, and thinking about writing and how to write more efficiently. Not saying any of the above comments will help anyone else do so, just that I invariably have a lot to say on the topic.
