I'd recommend using a range of reference tracks. Each song has for instance a loudness potential, as in some can go louder than others due to the sounds/arrangement. So what I usually use is a few reference tracks on the lower volume level (but still commercial) and then a few of the very louder ones. Same goes for brightness and other factors. Then I get the master I'm working on to fit somewhere within those general ranges.
Some guys use a really loud reference for a song that just won't get that loud (without negative effects) and end up trying to
crush the song and ruin it instead of settling for a less loud but still loud enough master.