In cases of a limited layout (i.e. a dungeon) the odds of not finding anything of interest in any room is slight; even if it's not combat (maybe there is a puzzle or trap to deal with). In wilderness or town adventures, the odds of PCs wandering into Dullsville get higher.
I have a solution I like to use: I just move the set pieces around on the fly. Go west instead of east, then I just move the haunted farmhouse (or whatever) into their path. In fact, for my own games, I've actually stopped placing things on the overland maps until after the encounter happens.
This can be seen more than one way, of course. On the one hand, it allows a level of player agency where you let them decide how to proceed ("We cross the river" vs "We stay away from the water"). Players like to feel like their choices matter. Otherwise why even ask them? On the other hand, if the trap or the monster is going to end up in their path anyway, what's the point?
My answer is that the set piece(s) aren't the only thing that the PCs might encounter. Also, once a feature like a town or dungeon or temple, etc. is encountered, then it is fixed on the map. I'm not going to move it about in their way– geographically or narratively. Unless the nature of the encounter is mobile (caravan or some such).
It's the same with time. If the players have no knowledge of a timetable to work with, then they will arrive when it is dramatically appropriate. On the other hand, if they are given a deadline to work with ("The scrolls say the hellgate opens at sunset!") and they choose to wait or are delayed, then so be it.
All this flies in the face of sandbox lovers but I have to admit, over the years I've played too many pure sandboxes that fell flat when it came to pacing and excitement to worry too much about it.
