I typically route two voices to their own outputs and process them with external fx. I like the internal fx and what you can do with parameters locks and performance mode, but using my own fx can be an awful lot of fun too (Meris Polymoon, Strymon Night Sky, Death By Audio Rooms, Catalinbread Montivillian et al) and opens up a lot of sonic possibilities. Having said that, I usually leave the other two voices on the mains and use them with the internal fx, with the option of adding a second voice to a track. I feel I get the most out of it this way.