One of the key, and in my eyes, most important reson to use sends for effects (time based, reverbs, delays etc and not dynamic changes such as compression) is that you can create a tracks 'listening space' and makes it a much more coherent sounding song. For example if you are using multiple reverbs on multiple inserts, its not really going to 'glue' together.
Typically I will set up two and sometimes 3 reverbs, one short one, and a longer one. Ok, for the short one, I would use it on snare(s), and also percussion and more. the beauty of using sends is that you can literally send whatever you want to it, and how much you want it to be effected. I usually also send percussion/ hats to this short snare verb but at a really low level, so you dont notice its got any verb on there, but it does add something to the overall mix.
My second most important use for sends is that your original audio is unaffected... and so you can group bits together in a buss for further processing, while still having reverb, but its not in that drums grouped and compressed. It keeps itself completely seperate. This is really useful for paralell compression, as you still get snare 'verb, but you dont hear it pumping away with the compression. towards the middle of making a track, i tend to buss each type of instrument (basses, drums, leads) together so i have total control of the overall mix. This is ware you can also buss all the send types together, like all reverbs, all delays, all flangers etc.. so you can control the global dry/wet of all the effects.
Now some people say they prefer using inserts for things like snare reverbs... but think to yourself why? its much more flexible, CPU saving and dynamic to use sends.
Post your own pro's / Con's here in this thread, and explain why you work a certain way. and i wont take "because im lazy i do it this way" for an answer
)