When using 32bit floating files as the default Cubase internal format, which are virtually impossible to clip, gain-staging isn’t as critical, except for where the DAW interfaces to fixed integer sources and targets, which usually means with:
a) Inputs to the ADCs of your audio hardware, usually 24bit integer.
b) Outputs to the DACs of your audio hardware, usually 24bit integer.
c) Rendering to fixed bit-depth audio formats, like for 16bit WAV files.
d) Inputs to some level-dependent FX like compressors, maximisers and ditherers.
e) Output from ditherers and maximisers, which should not be changed at all.
f) Outputs feeding broadcast equipment.
For these, gain levels need to be optimised to use the most of the bandwidth without clipping. At any other stage, gain staging helps the meters give a good idea of what the signal is doing at that point, because meters slamming into max scale or hardly registering do not help you to make finer level adjustments when required, or track problems.
Note that all these calibration ‘standards’ are rules-of-thumb generic setups, and do NOT guarantee that any particular signal will not go full scale. They just give some predictability, but which one you use will depend upon whether its required (as for broadcast), or what suits the genre(s) you do best.
However, if using fixed integer internal formats, gain staging is CRITICAL to avoiding clipping, as every audio track is a potential overload point when it is being recorded.
Note that Cubase uses 32bit floating internally, so it is only at points where audio files are created (recording or in-place rendering) that levels are critical.
When doing pre-recording level checks, note that for acoustic sources, like vocals or instruments, the levels encountered once the artists get into the performance while recording can be substantially above those from the artists in the more subdued level-check phase. Make sure you get them to do the level checks with a bit more gusto.