Sente or gote
The normal reasoning is this:
- the play may have a follow-up rather than being dead-end;
- the follow-up may be quantifiably better than anything else on the board;
- if so, the play should be considered sente.
The other steps are less certain. In the middle game, however hard one tries, it may not be possible to count at all accurately a follow-up. One has to apply some sort of triage: divide plays into clearly big, clearly small, and medium, and in case of several large plays apply that division again. Other things being equal, one prefers the open-ended big play to one which can be counted more precisely.
The old saying is: the play is sente if you hope the opponent doesn't answer it, and gote if you hope the opponent does.
In the endgame matters are clearer, once the macroendgame has passed. There one can discuss reverse sente in general terms and special cases, with numerical analyses. The idea of ambiguous play enters, showing that sente/gote can never be an absolute distinction.