I never 'normalize' per se. I add 'make up gain' to bring the maximum peak to -0.2dBFS.
I do, however, generally use some light compression when mixing more than a stereo AUD feed. Light compression + 'make up gain' is the same thing that 'normalization' does, with the main difference being that 'normalization' uses an algorithm to compute the dynamics processing and I prefer the specifc control of dialing in the compression the way I like it.
For a stereo AUD feed, I only add 'make up gain' and preserve the dynamic range as it is.