Normalisation finds the largest peak on a wav, works out how much it can increase the signal by at that point to put sound level up to 0db then uses that value over the rest of the signal. So say most of your audio was 0.25 of the full volume, with peaks at 0.5, it would effectively double the height of the whole signal so the peak would be at 1 and the main part would be at 0.5. It's a proportional increase across the whole signal.

If volume maximiser works how I think it works, by hard limiting the signal chopping off the peaks above a certain level then normalising it that increases the RMS power of the signal making it sound louder which could argubly make your mix sound tighter. However you are now introducing distortion into the signal in the form of the clipped peaks.