Reading this paper I'm amazed at the increased efficacy of some of the newer SSRI's despite not having a novel mechanism of action.

There are huge differences in the mechanism of action, quantitatively speaking, even within a class of antidepressants. As a particularly striking example, you're allowed to call your drug an SNRI as long as it has any detectable N effect at all -- even if the N effect is too small to practically make any sort of difference, and the drug is practically an SSRI.

Don't remember well enough to cite exactly but this may have been it: https://link.springer.com/article/10.1007/s11920-013-0370-7?