I went through arxiv.org primer and some questions at MO and here about arxiv. There are many of them so I could have missed answers to some of my questions. My apologies in that case.
I have never used arxiv before (not that I would publish many papers of interest). I am sure there are many people experienced with using arxiv at this forum. So here are some possible advantages I could see in using arxiv (feel free to add any other advantages of using arxiv). And I also have some question to these points, mostly to the second one.
Obviously, one advantage is availability of the paper. (IMHO This can be also obtained by putting paper on my own website, although the obvious drawback is that less people will stumble upon it. But I avoid intricacies of using arxiv - I guess there are some things you have to learn)
Obtaining feedback before publishing the paper. (I am not quite sure about it, so perhaps this is where some experienced arxiv users could give a hand.) If you put a paper on arxiv.org do you really get some feedback from people who read it? Should I leave some time between putting a paper on arxiv and submitting it to a journal - so that I can incorporate some feedback which I find useful? (I'll give a few examples. The results from my very first paper based on my M.Sc. thesis - or at least one half of them - were obtained before in a M.Sc. thesis at another university. However they were not published in a journal and I found a paper where that problem was mentioned as an open question. Putting it on arxiv would increase the probability that someone who knew the mentioned MSc thesis would see my paper and let me know. Similar case: Some time ago I have seen a paper proving an old result, the authors obviously were not aware of it. Now there is a comment "This paper is withdrawn: I have learned that the result is known.")
Communicating result to other mathematicians sooner. (It might take up to two years before a paper will be accepted in a journal. Putting preprint on my website and arxiv is better than nothing - although I am putting out a version that has not gone through peer-review.)
Publishing material different than papers. (I believe I have seen some PhD thesis and maybe even book drafts at arxiv. There is a question what can be published. I have seen at arxiv primer that: "Material submitted to arXiv is expected to be of interest, relevance, and value to those disciplines. Submissions are reviewed by expert moderators.")
EDIT: Some of the comments suggest that it is unclear what I am asking. I am interested specifically about the feedback to papers published at arxiv (which I mentioned in 2 and which is already addressed in Jim Conant's answer) and more generally, any advice to an arxiv-newbie is welcome.