To me parallel processing is just mixing an affected signal with the original. I don't use parallel compression that often, but I do some parallel processing with basses and distortion a lot. Using sends is very similar to this process.
A quite popular technique among pros in regards to parallel compression, is multi-stage compression, e.g., two stage compression. The configuration is like this:
A) Track 1 (dry source)
B) Track 2 (comp 1 in parallel)
C) Track 3 (comp 2 in parallel)
A -> B
A -> C
B -> C
C = result
So you get two signals summed before the secondary compressor, where one of the signals is already compressed by another compressor before it hits the secondary compressor. This you can in turn use creatively, e.g. you could make A only be a frequency band, let's say the low mids of the snare. This means that you get to control the compression of the soft and loud low mid playing, separately. So instead of losing the body of the snare in the chorus, you are now able to adjust the secondary compressor so that the peaks of the dry signal that are increasing on the louder playing are making the overall compression a bit "loser"/"less tight", which really in many cases can sound very sweet because the tone is kept more natural across the whole velocity range yet under dynamic control. The issue with most bad sounding mixes out there is due to compressors that become signal eaters on more loud song sections, commonly much increased due to automating on the wrong side of the compressor. It is often sweet to losen the compressor grip a bit on the louder secctions to allow the power of the playing shine through, by not getting locked up by the compressors, especially when it comes to drums. I like to use two stage compression on pretty much all scopes, I like to do it that way because it makes it easier to not overdo the compression and it allows me to control the dynamics of the whole velocity range, so that it does not perform well only on some loudness levels. This helps also to balance the frequency bands across various loudness levels. It is common that snares turn into gun shots in the chorus, because of the upper mid range (in conjunction with a poor monitoring process). In this case it might help to have two stage compression on the low mids, and single stage compression on the upper mids, essentially what then happens is that the body of the snare takes over in the chorus, which can sound big - the body frequencies open up and the splash frequencies are silenced. A gain control after that will boost the effect even more, so now I have taught you how you can make the snare loud, but not unpleasant to the ears. The louder the playing, the more body and less splash. So in mixing, it's a lot about paying attention to the sound of these in the softer sections, separately from the sound of these in the louder sections. Sweet mixes are commonly sweet because of how the dynamics behave across the song. If you have heard the effect of compressors in the verses, the ears expect that to continue in the same way into the chorus, but when it opens up instead it sort of becomes a release. Big mixes are big because the compressors allow them to be big. You can do the same with gain, it's just that then you get harsh sounding transients instead... So you want big fat sound without transient harshness. And you do not really have to split tracks if you don't want to, it can be enough to just implement a two-stage multiband compression chain in parallel. Because sometimes it is more important to really control the dynamics across the mix quickly, without all of the time it takes to set up the fx chains. You want to quickly jump at it and get the sound you want as quickly as possible while your ears are fresh. It can also pay off to work on fairly big scopes to reduce the impacts of phase, many top mixers do it that way for that very reason. Be aware of the impacts on the stereo image though and try to find ways of doing it this way in a stereo friendly way.
Also remember that you can work with expanders in parallel too, and other effects of course...
Music creation is tricky in the sense that its dimensions are polarized, this polarization causes paradoxes that the engineers easily get locked into. Boosting the high end creates clearity so that is good, reducing the high end creates softness, so that is also good. On the contrary, boosting the high can create harshness, reducing the high end can cause lack of clearity. Warm low frequencies are accessed in the high end, how can that be. It is a paradox. Do you see what I mean. So it is about effectively dealing with all of these paradoxes and to do so you have to be open to this kind of complexity and deal with it according to how it works rather than according to how the brain wants it to work.
Some of this is stuff that you don't do effectively during the production and recording process. You should, but it takes a great mixing engineer driven by great monitoring to do the fine balancing towards that very top quality of the mix spectrum. Hence this is a process that belongs primarily to mixing, you should set it up in recording and you should do it the best you can there, but the mixing engineer ensures it ends up at full balance with certain emphasis on certain colors, based on the mixing engineer's taste and monitoring environment. (that should be as good, and better - even better than the recording monitoring environment)