- Make the players invested in the history of the world. This depends on the system you're using. However, generally, you want the players to come up with a backstory, yes? Well, nudge them to tie their character's backstory in with the rest of the world. That way, when something happens in the world, they'll be emotionally invested in its outcome. Hence, direction.
- Design the world so that things happen. What it sounds like to me is that your worlds have a lot happening, but not much that directly concerns the players. If an optimally lazy person wouldn't be motivated to do something about it, you can't expect your (motivated) players to respond to it, either. There are a couple tricks to this, and I'm going to go into more detail below.
In order to create a world that engages the characters, Things Need to Happen. But even then, players probably won't respond to much unless they need to, so the Right Things Need to Happen. What constitutes a Right Thing?
- The Things have tangible, immediate impacts on the players. If they're motivated people, and they hear about a mysterious ghoul in the town next over who might enter their town, they probably won't do anything about it. They'll defer to that town to handle it, at least until the players absolutely need to respond to it. In this way, if the effect isn't tangible ("some of the peasants might get ill") or immediate ("if you don't head this off now, there will be serious problems in, oh, a decade or two"), they won't be engaged.
The Things must be something the player characters are invested in. If Matilda the Barkeep's husband has been captured the The Allthing of Evil, are you really going to care enough to venture to the Veil of Certain Peril and retrieve him from the Claws of Destiny? Probably not. Especially if you don't know Matilda, her husband, or don't really care about the town.
It is your responsibility to write the game for the players, as much as it is the players' responsibility to play your game in a motivated fashion. It sounds as if your players are definitely motivated, and they're looking for something engaging to do, but also that the game tasks that exist aren't really planned. Even though this is a sandbox game, the things that happen still need to affect your players.
"So, I've thought of something(s) relatively open-ended that their characters would be interested in tackling. What next?"
Think about it some more. Then think about it again. Then continue to think about it while you're playing the game. You need to consider in particular what the NPCs are motivated by, who they know, what they know, and what they plan to do about it.
Now, you don't need to know this for every NPC. However, the Special NPCs (SNPC, I call them, but that's a LARP habit), who drive the plot forward, create new twists, and are the NPCs who truly engage the players, need to have well-defined short-term and long-term goals, as well as basic personality sketches. Every NPC should have something(s) they can, with some difficulty, overcome, that prevents them from reaching these goals. These things should tie in to other players/NPCs/SNPCs.*
Throll has a long-term goal: he wants the Axe of Beldor, but unfortunately, it's buried beneath the Tombs of King Calthor. That wouldn't normally be a problem, except the Tombs of King Calthor happen to be underneath the city; in order to get to them, he must destroy several houses. Maybe of some people your PCs know.
He wants to use its magical power to destroy the peasants' rebellion. He has no qualms doing this.
When I say every NPC, I mean (pretty much) every NPC (though regular NPCs don't necessarily need to tie into the characters).
Matilda the Barkeep has a long term goal: She wants to build a second inn across the city. Unfortunately, she doesn't have the funds for it currently, as she was robbed by a mysterious bandit two days ago! Oh no!
Your players may or may not be engaged by this. If they're not, more options are needed. In a sandbox game, more options should pretty much already be there.
There aren't many SNPCs in a short game. Their motives are thick, and they are often (but not always) weak people. You only need a handful to make an interesting conflict. Give a bunch of twisted people power, and their scheming will stretch across cities.
The other type of NPC is the regular NPC, who don't get a fancy title or anything. They're just NPCs. These people need to have plans and know others as well, but theirs can be more abstract or simple. Your barkeep Matilda, for instance, could have a short-term goal of "I need to get more liquor." and a long-term goal of "This place needs some expansion." This alone will give your NPC more flavor, as they will be where they are for a concrete reason.
Just these two things will drastically improve your campaigns. To summarize:
- Get your characters to build a solid character background.
- Design the world so that Things happen.
- Make said Things concern the players, in a way which allows them to engage the world.
- Give your SNPCs complex and concrete motives. Give NPCs some minor motives on the fly. Some minor NPCs might know someone helpful.
- Let your players guide themselves through your story, and be free to modify it on the fly to match what they appear to follow.
This approach does make the game less of a pure sandbox, but will make it a more engaging sandbox, if a sandbox truly is the style of game your players appreciate.
*Side note: The difficulty and length of the game is often very closely related to how difficult the goals of the NPCs are to attain. Goals which are complex and will take a while to implement will result in a longer game; shorter goals, like Throll's, will make the game significantly shorter.