Great topic! Love it!
As you might know I have studied audio engineering and I remember that one of my educators said: “Forget the Inserts or Sends…”, what he meant, as an engineer you need to figure out, what works best in your situation. There are a lot of people out there, who are saying, that Delays and Reverbs should ALWAYS be Send-FX-Returns. However they’ve just learned it, “You need to do it ONLY like that…!”, (like 2+2=4), but actually never “explored” other options for themselves. It’s not bad or whatever, it’s not “engineering” really. Engineering is trying new crazy things, fail, go back, try again until you are happy with your result. If you are still unsatisfied, relax, think about what you want to achieve, and try to build it from scratch. (For this you need to know the basics of course). Basically what people need to do at all other “creative” jobs as well. For me it’s really boring at some point, if I do the same and same stuff again, I need to feed my brain with new things, or I just stop thinking and just do what I already did. And then it becomes a bad routine in that sense. You will never grow your creative portfolio, if you take engineering as a craft you only learned once and for all. (My little excursion on engineering and creativity in general…and now back to the topic )
In what situations I like to use Insert FX: For example I have a pretty dry instrument, a flute which was recorded with a closed mic, but has no room. I might put a short delay or reverb, (depends on the color and the hearing result), and insert directly on the instrument track. Because I used the reverb on the original channel I can position my signal in the room as I would like to. If my flute is mono, so my reverb will be mono as well. Make sense, because the ear identifies mono-fx much better, as they are “concentrated”. You can EQ and automate your channel as you like, so the original channel plays with the music and acts like a real instrument, so do all the insert fx.
If you have a signal which acts just as an FX, you don’t need to send it somewhere else. Just through a reverb in the insert. As long as you don’t want to “parallel” anything, you don’t need your sends.
When I use sends: Sends are copies of your original signals. With your “send-knob” you just adjust how much of the original channel is going to the return, where all your FX are used. That said, it makes sense to treat your Return as an individual instrument. If you send your brass-section to it’s Return, you can manipulate your reverb and EQ your reverb, compress your reverb, delay your reverb, duck your reverb, saturate your reverb, etc. but you just touch the copy, and not the original, so you can always dial in the fader, or even use crazy automation rides, and so forth. Basically you always have much more options, as you do it in parallel. With inserts it’s not possible, until you automate your dry-wet-knobs (not every plugin has it).
You can say that theoretically your send-return is your “dry-wet”-knob.
Keep in mind, that it makes sense to use more send-returns, as first you are more flexible, and seconds you use less of your CPU, as you can always send more signal into one return. When I use insert FX, I usually print it.