I have found that different areas of research have their own style and implied requirements. It is always good to start with a literature review. Comparing yourself to previous work or at least acknowledging related work is universally necessary (at least in every good publication I've seen).

My process for any kind of writing is to make an outline and then keep adding detail to the outline until the only thing left to do is convert bullet points to sentences. I also usually allocate at least 5 or 6 weeks to actually writing the paper, not including the time to assess ideas and collect and analyze data.