http://www.scientificamerican.com/article/antidepressants-do...
http://slatestarcodex.com/2014/07/07/ssris-much-more-than-yo...
...which also sort of goes along with this overview of depression treatment in general:
http://slatestarcodex.com/2014/06/16/things-that-sometimes-h...
The relevant upshot of that first link includes:
(1) Everybody agrees SSRIs are at least somewhat better than placebo
(2) Placebos really are unreasonably effective compared to non-treatment, so even merely being as effective as placebo is a huge benefit, not to mention being somewhat better than that.
(3) What constitutes a "clinically significant" improvement (even relative to placebo) turns out to be kind of arbitrary. As in, a drug that misses that arbitrary goal might still be hugely beneficial to people. (especially if you compare it to non-treatment)
(4) Glomming all SSRIs together into one big lump to reach conclusions about the group combines some that work pretty well with some that are known to not work well at all, which can bring down the average effectiveness of the group to below that arbitrary "clinically significant relative to placebo" mark even though individual SSRIs might on their own be more likely to make the cut.