I've been asked to say something concrete and practical about how to write a conclusion. Some people, it seems, have gotten all the way to the end of their PhD studies without ever having to write one! So let me try to say some pretty standard and elementary things about how to go about it.
Anne Huff (1999: 92) has more or less said it all, actually. She nicely summarises the main pitfall of the introduction-body-conclusion model, for example, and thereby also the standard (though specious) argument against using that model:
Scholars have to be careful that they don't bore intelligent readers by repeating information from the introduction in the body of the paper and then reiterating it once again in the conclusion. The strategy of repetition also forfeits the opportunity to deliver as much information as possible to skimming readers, enticing them to read the whole paper because their time was well spent looking at a brief amount of material.
Notice that the conclusion should be written with two kinds of readers in mind: the intelligent ones and those who are skimming your text. These may of course be one and the same person. Smart readers want to know what they will come to believe if the body of your paper is convincing. They should be able to learn this simply from reading your conclusion.
So it should be able to stand alone, as Huff points out. Keep in mind that "readers who have worked through your paper typically need to be pulled back to the big picture with a conclusion." You achieve this by writing "the most assertive statement of your paper's benefits that you can make" and then indicating a way forward: "add some last thought that enhances your work's significance" (Huff 1993: 92).
That last thought should be the only new piece of information in your conclusion. You should not introduce new textual references or factual claims. The last thought should be like salt in a rich sauce: it should sharpen the edges of your argument, crystalize the structure of your materials.
One way to do this is as follows. In order to avoid adding anything new, you construct your conclusion exclusively by copying and pasting sentences from the body (not the introduction) of the paper. You next introduce the salty new thought that brings it all together in a significant way, and then rework all the copied sentences so as to add up to that new thought (again without adding any new information).
Doing it this way, however, is possible only if you have left the conclusion to the end, and I don't actually recommend that. Like Huff, I recommend writing the conclusion first. Kafka concurs.