I feel putting your compression in the group channel is not the right thing to do. Instead of "seeing" a vocal track and working on it's dynamics the compressor sees a sum of signals and compresses accordingly, which means that when one track is loud at one moment, all the other tracks are compressed as well that moment. Even if they dont need it. To me that's against the purpose: You want louder parts of one track compressed, so that the other tracks that arent as loud dont get lost. Or am I wrong here?