Many of us know it's importance in mixing, but seldom do we discuss it in the context of mastering.
Gain structure is important here too.
See the below screenshot.
I have a gain utility at the very beginning of my chain (Gain 1), another after my compressor (Gain 2), and lastly one after my limiter (Gain 3). Let's break those down:
Gain 1: This is for managing my own peak headroom. People often say that mastering engineers need 6dB of peak headroom, but that's not an absolute. Mastering engineers will manage the gain going into the chain anyway, which is why I have placed a gain stage at the beginning. If I (or the next plugin) need more headroom, I can simply apply negative gain to bring the peak down.
Gain 2: This is to manage the gain after my compressor. As I'll likely dial down Gain 1 with a view of giving my compressor's threshold more headroom, it's likely I'll need more gain afterwards to bring the volume back up to my mastering level. This also allows me to rely less on the limiter's built in gain. Whilst not a technical difference, it does give me the freedom to toggle the limiter and judge its effect objectively without a change in loudness. This is known as an equal loudness workflow.
And finally, Gain 3:
Firstly I should explain that the limiter there is set to a True Peak ceiling of 0 dBFS. Gain 3 is set to -1 dB. This results in a True Peak level of -1 dBFS.
Before I print the master, I activate Gain 3 which turns the signal down by 1 dB. This gives the master some headroom to mitigate the risk of clipping when encoded to lossy formats (e.g. MP3). However, I will at times toggle the bypass to hear what that extra dB of difference sounds like.
Hope this helps. Please ask questions.