What's your approach to managing dynamic range in a mix to ensure both clarity and impact without over-compressing?

Personally I prefer to do a a little bit of compression, more often. So rather than smashing something with an 8:1 ratio, I'll do a little bit of compression as the vocal comes in, a little more on the vocal bus and then of course the mastering compression.

Same with other instruments, a little bit of compression more often is less noticeable than one compressor really working hard.

I also find theres a balance thing in the audio people often over look. Before hitting your mastering button, the vocals, drums and bass probably want to feel a little too loud in the mix so that when you compress everything, they're still the dominant items.

Simple, dont compress unless you have to. 

Make use of clippers, envelope shapers, volume automation and saturation before you reach out for the compressor. 

When you have to compress, use a serial-light compression approach, meaning stack compressors but make sure they're only working minimally, like maybe 2-3 dbs of gain reduction at most. 

Compression isnt added, its multiplied. so 2:1 ratio followed by a 4:1 ratio is actually 8:1 ratio and will result in that much gain reduction. Another 2:1 after that means you get 16:1 reduction. 

However stacking them will let you get you less noticable compression. So instead of using 1 16:1 ratio compression doing 20+ dbs of gain reduction use 3 which are working minimally. 

Volume automatisation is the key. Don't use compression with a ratio of more than 1.5:1 on the master bus. Don't forget that compression sums up so, for example, your vocal with the compression 4:1 will also go through your master bus compressor and a limiter which is also a compressor.