You're trying to give them options by withholding options. You want them to explore the verisimilitude of the world by not describing the world. I can understand this makes sense in your head, but it doesn't at the table. Particularly with new players.
They're new players, so they need to get their bearings. They may have no idea what kind of options there are, what kind of choices make sense. They take your guidance and follow up on that. If you don't want them to follow your lead, give them other stuff to follow.
If that sounds like totally not the game you want to run, don't worry. They will eventually get their bearings, and start making their own plans, and before you know it, you'll be cursing them for constantly derailing your plans. I hope when that time comes, you'll remember that that's what you wanted.
Anyway, there's a few different approaches you could take:
Giving them a few starting options, but hint that there's more
When they want to infiltrate a city, you could tell them:
GM: "The way into the city passes through a massive gatehouse. This is where normal people enter the city, but you might to run into guards who ask questions. Is that the way you want to take or do you want to look for a more subtle entry into the city?"
Players: "Er, let's take the subtle entry."
GM: "Alright. What are you thinking of?"
Players: "I don't know. What other ways are there?"
GM: "Oh, lots. But which are the good ones? Do you want to climb over the wall, perhaps?"
Player 1: "Yeah, I'm good at climbing. Let's do that."
Player 2: "Wait, won't there be guards on the walls too? Climbing seems more conspicuous than simply taking the road everybody else takes."
etc.
Sewers are also something not every starting player thinks of. Mentioning it explicitly at least once, possibly a few times, may get them to consider it as an option. Or maybe they think it'll be locked with a grate.
Or, as they walk on the road towards the main gate, you could simply describe a bit of crumbling wall with some sturdy vines and a sewer exit, as well as some other stuff that makes the city feel real, but isn't necessarily a potential entry. Describing the environment is always good. It brings the whole thing to life. There may be plenty of other ways into the city, but simply by walking there, they already notice a few.
The same goes for the shops in the city. Even if they head straight for the tower, they're bound to walk through a number of streets and see a variety of shops and inns there. Describe them, even briefly.
There are different ways to handle this. You could simply say: "On your way to the center of town, you pass an inn with a big crowd outside. A bit further down the street you see a weaponsmith's that's boarded up. As you round a corner, you see some people kicking in a door of a shop that has above the door a sign shaped like a vial or something."
Or you could hide it in the atmospheric description of the city. Describe the smells of the tanning district they pass through, with the leatherworker selling a variety of leather equipment and armor. Later on, they get to the fancier more upperclass mercantile district, where the merchants live, though you can also find alchemist's shops here. If you ever need potions, this is the place.
But these methods still have the problem that you're constantly deciding what they see, and they might believe you have a specific reason to show them exactly that and not something else. You're still leading them. There is a great way around that; an old tradition in RPGs, that's unfortunately surprisingly controversial:
Big Random Tables
The idea is simple: make some big tables with random events. Make every event cool. And make it clear that you are rolling on a table. That last part is vital.
Random tables, particularly random monsters, got a bad reputation, because a lot of people feel that the GM should simply present awesome, relevant encounters, and not risk randomly rolled crappy irrelevant encounters. And admittedly, a lot of random tables are crap. But there are two big things that random tables accomplish, that the GM picking the most interesting relevant encounter doesn't accomplish:
- The GM is not leading the players anymore; you give this control out of your hands, and if you can't give it to the players, you give it to a die roll. At least you're not dictating the course of the adventure anymore; you don't know what's going to happen any more than the players do.
- More importantly: if the players know you're rolling on a table, they also know that there is more on that table. They will want to know what. The existence of the table already gives them reason to explore. There's a big world of people and events and stuff going on, and much of it may be irrelevant, but hopefully it's also all cool in its own way.
Not everything that happens to them is automatically relevant, so suddenly, they have to decide what they respond to. Ideally, every entry on the table has its own story (which may or may not tie into the greater story), and if the players see the story, they may do something with it. If they don't see the story, they'll probably ignore it, which is also totally fine. By making that decision, the players are suddenly deciding what they want the game to be about.
And they can afford to ignore some events, because they already know it's not vital, because there was never a guarantee that it would be rolled. But it could still be interesting, relevant, or offer a cool opportunity, and they'd be crazy to ignore that.
Designing a good table of random stuff
Note that a table can contain anything. In the past, D&D often had tables of random monsters you could encounter in a dungeon, but it could just as easily be about shops and other events they encounter on a walk through the city, or rumours they hear in a bar, or even possible entries into the city.
A good table entry has a story. Too many random monster tables said nothing more than "1d4+1 Orcs" or "1d3 merchants". That sucks. Why are the Orcs there? What kind of merchants are they and where are they going or coming from? What makes them or their situation different from other merchants or orcs? The best table entries tie into and modify the greater story. Perhaps they meet an important NPC, run into a vital clue, see the handiwork of the evil henchmen, etc. Whether they have or haven't had some encounter may have impact on how other events unfold. Maybe different encounters provide different bonus clues that may lead to different approaches to the main problem.
But there's another technique that could also help you:
Ask the players about the world
The traditional way of roleplaying is that the GM is the final arbiter of every fact in the world, and the players are constantly asking about them. Or if they don't, they simply won't know much about the world, which would be a shame.
Dungeon World is a game system that turned this around. Instead of answering questions, the GM would constantly ask the players about the world, and the players' answer would define a little bit about the world. The GM would't ask about the main plot line of course (unless you want to go there, which you might), but you could ask: "Other than the main gate, what other ways into the city are there?" or: "On your way to the tower, what kind of shops do you see?" You make the players co-authors of the world, and that forces them to turn on their creativity, and not sit passively waiting for you to spoon feed them their options.
And that's really what this is about, isn't it? You want them to think creatively, so make them think, create and decide, rather than hoping passively that they might think of something on their own.
