Jonathan and Andrew recommend against "signposting" in your writing. This recommendation is in line with their cultivation of the "classic" style, as presented in Thomas & Turner's Clear and Simple as the Truth (2011). I haven't read the book, but Princeton University Press tells us that
In classic style, the motive is truth, the purpose is presentation, the reader and writer are intellectual equals, and the occasion is informal.
My approach to academic writing endorses all but the last of these; that is, I think academic writers should present truths to peers, but I don't think formalities are entirely out of place. This may well be because I work mainly with writers in the social sciences. The style of writing in the humanities is probably less formal, more classic.
Among the formal constraints of social science writing are the need to present your theory and your method. And the need to state a clear thesis and draw some theoretical or practical implications from it. Another is the need to summarize the argument, first, briefly in an abstract, second, in the introduction and, third, in the conclusion. (These three summaries are directed at subtly different "audiences"—an issue I will speak to in a later post.) The constraints are formal in the sense that they are imposed independent of considerations of content. The reader expects to find a theory section, a methods section, etc. The reader has certain expectations of what the the introduction and conclusion will tell us.
This allows the (experienced) reader of social science to read a paper very selectively and therefore survey its contents very quickly. The classic style, by contrast, presumes that the reader will read the text from start to finish. This presumption, I want to emphasize, is very good for your style. If you write a section on the presumption that most readers (or just bored ones) will simply skip over it, you're liable to lower your standards while editing it too. What I suggest therefore is writing a paper that has a certain kind of surface structure (including signposting, e.g., "In this paper, I show that...", "This section will shift our attention to...", "I will now draw out a number of implications...") but that would make sense without that structure. The best of both worlds, in a sense.
In my heart, I agree with Andrew and Jonathan. Whenever you have to talk about your argument, the "paper-ness of the paper" as Andrew put it, you are taking some of the intensity out of the presentation. But that's also what formalities are for. In business and legal writing, as in social life quite generally, various institutions guide us through what would otherwise be complex intellectual and emotional situations in a few simple moves, recognized by all as well-intentioned, if often not very interesting, attempts to satisfy a range of interested parties. It is sometimes said that we resort to formal rules in order to "avoid misunderstandings". Sadly, but I think unavoidably, we also resort to them in order to avoid understanding each other too well. More constructively, we might say that they spare our readers the trouble of having to understand everything we think we know in order to engage with some of it. The formal constraints of academic writing allow us to share our thoughts with each other without having to share all of them. They let us examine each other's ideas one at a time, as Ezra Pound hoped we could.