Ok so I've been learning a lot about this whole mixing game. I've done some mixes for some university projects and I'm starting to get somewhat of a hang of it. However, there's one thing that I can't quite decide on.

I know that compression is used to reduce the dynamic range of a signal, and automation is used to control the faders (but can also be used on things like pan and effect parameters). I also see the clip gain option in pro tools and I understand that.

Where I'm having trouble is knowing which one to use. For example, I have the stems of one of my bands songs we recorded. Essentially the song has a quiet intro and verse, a loud chorus and solo, and then it repeats this cycle. We recorded the guitars and bass for the soft and loud parts at different gain levels to get the best possible signal. However, the vocals and drums were done at the one gain level because there aren't any gaps in the arrangement and they also transition from soft to loud within a few beats. I'm not sure which order/method to use in this case.

A few things spring to mind:
1: Do I compress the drums and vocals, or do I automate them? Or do I use clip gain (only on the vocals)
2: How loud should I aI'm to make each section? Generally I leave about -6dB headroom on the master fader for anything loud, so should I bring up the soft sections until they are -6dB?