For me, I just isolate the portion that clips and use a gain change to drop it 1 db at a time until it stops tripping the clip-o-meter. I know that sometimes there is a random measure somewhere that is at a higher level of attenuation than the rest of the track.