I’m in a social science where papers are not quite as structured as some of those who have responded to ecogeofemme. Generally, once I’ve identified what I'm writing about, even if I know what I want to say, there’s still some reading I need to do to situate my work within a couple of sets of literatures. Sometimes, these are ones that I’m closely familiar with, but I generally still need to re-read to remind myself and also to refine my own argument.
While I’m reading the sources, I take notes on quotes that I might end up using, as well as picking up on other sources, and especially noting down points of interest, contention etc. As I’m brushing up on the literature related to my topic, I’ll periodically type up the notes in a form that I keep revising until it looks like a literature review. I start with all the notes on everything, then I go through and highlight the parts that are relevant to my argument. Even if I had an argument, it inevitably becomes more clear and easily expressed by the time I’ve done all the reading.
I don’t make an outline until after all this is done, because otherwise it just ends up getting scrapped. Once the literature review is written, and I can see what the article is contributing to the literature, the logic of presenting my own argument and supporting points flow easily from there. Only after I have a solid middle section draft do I start on the introduction and conclusion. And that's probably not even halfway there, because the real beginning is revising.