Gain staging, as it exists in the analog world, is indeed unnecessary in a floating point digital audio system such as Reason. This is something a LOT of us have been saying for years now!
Though I was consulted for Ryan's video (above), I still felt he didn't cover everything important and maybe made things a little more complicated than they needed to be. So I made this video simply showing my personal workflow in Reason, because this question comes up again and again! IMO all that's necessary is to keep ALL signals at all points within a user selected range of peak (not VU or RMS) levels.
Questions that may come up:
Why use Peak levels and not VU or RMS?
Because the goal is to keep your outputs from clipping, and it's the peaks that clip.
Why leave headroom and not just normalize all signals right up to 0 dBFS?
Because you are combining multiple signals when you mix, and if even two signals both hit 0 dBFS at the same time you'll already be clipping by 6 dB (two identical signals summed will be 6 dB hotter than either signal on it's own).
Why worry about levels at all?
Because it will make your life easier - really! If you keep all your signals at the same level (leaving headroom), it makes setting compressors/gates, and saturation/distortion devices quicker and easier as they always "see" the same level at the input. The end goal here is to keep levels at a point where you can stop worrying about clipping and spend more time mixing (and less time fixing).
What peak level is best?
That's up to you - if you mix 8-16 tracks max in your productions you can use a hotter reference level. If you tend to use upwards of 32 or more tracks, a lower reference level may be necessary. The Reason User Manual suggestions recording audio to peak around -12 dBFS. The Reason Factory Sound Bank (newer patches) are set to peak around -12 dBFS in average use. I find that level to be perfect for my production style - I never worry about clipping as my mixes always hit peaks below 0 dBFS by at least 3-6 dBFS before adding any inserts or mastering.
But just try it as a starting point and adjust from there if necessary for your own workflow.
What about those input knobs on the SSL?
I hardly used them in decades of work on a real SSL, and I hardly need them in Reason because of using the above method.
What about when you add EQ or a compressor - doesn't that change your reference level?
Yes, and that's why you need to use makeup gain or an output level control to restore the original level. This also makes doing "before/after" comparisons easy (I always suggest "before/after" listening tests when adding processing).