When frequency comes into the picture is when the comp contains a crossover (as it sounds like NGU's comp does). The comp will apply compression in different amounts (or not at all) to each channel of the crossover, depending on the settings you choose. The main benefits of multi-band compression are:
There can be very strong low freq spikes which we don't "hear" as a high volume level. With normal full-range compression, a large spike in the low frequencies will trigger heavy compression, and the rest of your tone will get sucked down with it. Notes can seem to "disappear" in the mix that way. Compressing the lows more than the highs fixes that problem.
Most of the mwah of a fretless and the bite of a slapped fret (for example) are in the higher frequencies. Compressing the highs more can give you more sustain from a fretless, or allow you to really spank your fretted, without exaggerating the rest of your tone.
OTOH, some people feel that the audible artifacts that come with most crossovers are not worth the benefits described. Also, most of the problems and/or benefits described can be regulated or achieved with good playing technique.