I have quickly skimmed though the video. You should take it with a grain of salt because he is entirely relying on Spotify’s FAQ, which is outdated on several things (it still says that Spotify is using ReplayGain, for instance), and has some debatable information (like advising everyone to master at exactly -14 LUFS, as if it was a target that would apply to everything).
There is a lot at play with loudness normalisation, and this is changing constantly. We don’t need to know all the details at all times. Basically, if you make a great sounding master it will still sound great normalized. Dynamic range is left untouched in the vast majority of cases, so there really is no need to bother with too much detail when mastering, apart from the dBTP headroom (which is also explained in very simplified terms in the FAQ by the way).
Focusing on making the best possible sounding record is still the way to go.