Someone is always looking, the client. If I make it too quiet, they will complain. If I make it too loud, they will complain. This is why great communication is at least 50% of a master that both me AND the client can be proud of. There are no prescriptive numbers, and every track has a different loudness potential.
All else being equal, I will try to make things as loud as possible before audible degradation, but that only really applies to singles. The LUFS Integrated figures where that ends up could fall anywhere. Again, the numbers are not prescriptive or “targets”. With an album, all the tracks need to sound cohesive, and that doesn’t necessarily mean the same loudness. The track that needs to SOUND the loudest still has to have no audible distortion, and all the others need to be juggled around that.
Sorry to sound obtuse but that’s really how it is. I have no “personal preference” without another qualifier, such as “What’s your personal preference for well mixed, sparse, beaty Chillout with decent dynamic range”, which I would hazard would end up somewhere around -12 to -15 dB LUFS Integrated, etc.
Here’s a compilation I mastered for Interchill last year, of just such music, I think the average LUFS Integrated figure is around -14. AFAIK BC doesn’t change the volume, so should give you an idea:
I no longer use RMS as LUFS is just RMS with a better weighting curve to match more closely with human hearing. It’s not perfect though, which is why it’s still necessary to do everything by ear, and let the numbers fall where they may!