The key is to not neglect any phase of the game and to spend the most time on the phase that you are weakest in .
+1
This is true not only in chess, but for learning in general. Approach things as a whole -- and the quickest, most efficient gains are usually from spending time in the weakest area.
In general I would agree -- in fact this kind of philosophy is something I live for, but it also depends on how important the "weak area" is. I could be extremely bad at mating with queen and king vs king and rook, but since that will probably never occur in my games I don't necessarily need to work on it.
I knew you would make that comment.
Opening is important, what distinguish strong player from a duffer is the opening. If I did not give some time as a low rated player and played junk opening I would still be low rated; I seen players from my two chess clubs I play and players who play junk opening or avoid playing 1.e4 and standard lines ( Colle, London System, King's Indian Attack, Reti and finally English.) stay low rated, if are below 1800 elo you need to play 1.e4 so you can develop to become a better player; the reason 1.e4 is important is you get tactical positions and positional positions and others different pawn structures and different way to handle center breaks, for insteads the French opening, Sicilian, Petrov and Ruy Lopez ; each of these opening require specific planning to reach a desire goal, and that is a opening advantage. Your opening needs to be sound enough to play against Master.
Not entirely true...many GM's play those openings, especially the Engllish.
In fact, if my chessbase research is correct, there seems to be a secular trend away from e4 towards Nf3/c4 systems at the top levels. That doesn't have much bearing on what opening repertoire you need to get to 2000, but it's certainly an interesting development.