Master painter Ives Gammell, whose voice I heard while doing a tape transfer for a painter, former pupil of his, would say, when editing a painting in progress, first address the things that are "furthest wrong." Then work on the details. Don't do detail work first and then have to do broad strokes that could undo any of the precision work, for sure...
(too quiet?, too bright?, too much bass? Where? etc..., back off, worse? better? You know. Bob's your uncle...)
However, as I believe Bob Ohlsson once wrote, every once in a while, a mastering session only reveals that the mix is twisted out of repair, due to acoustic and/or electrical idiosyncracies of the mix room. Sometimes, it can't be made to translate outside of the room it was mixed in without severely compromising some key aspect of the mix. Might be the engineer, or the day, or the key of the song and instruments used that excited problem modes, etc... Did an intern accidentally mess with a cross-over or polarity of device on some channel or other...? Also, and this is my opinion, now...most recordings are destined for the ash can of Life, since only the rare treat can be (considered) great, and all else _has_ to be common and/or plain bad. This is not your fault. This is Science.
Just fix what's still wrong after you make it louder (if'n it "needs louder.") If you eq before you invoke a comp or limiter, you can get thrown off course once the loads change and the side chain starts reacting unless you're mindful about the auspicious present. Keep it simple. Live in the moment. Judge what you're hearing now, not then. You might need to follow up with more eq after a comp, even though you might have eq'd going into it, too...
Also, while haste sometimes makes waste, if you work too long on a song, you could get used to a bad sound and hear it as ok, only later to revisit and hear that an obviously needed adjustment had been omitted (or that something was stepped on too hard).
FECCPR (Fresh ears consult comparable program routinely.
)
Andrew