When using this tool, the aim is to attract the player's attention to a certain location. This is often used in linear level designs with complicated rooms to make it clear where the player has to move in order to continue. It can also be used in less linear levels to highlight important areas of interaction -- such as places that the player will revisit often.
Essentially, the idea behind this is to make this target area look more important and thus interesting to the player, so that he moves closer. This can be done quite subtly, so that the player is unconsciously pulled towards where you want him to be. Of course it can also be done using more blatant means which will most likely cause the player to respond more consciously. As you can imagine, there's quite a number of ways to go about this, so I'll try and list a couple examples below.
If we want to attract the player, we first want to catch his eye. Here, painting techniques help us a lot. After all, painting has been manipulating and steering the human eye for millennia. What's true for paintings is true for levels: Visual contrast is a good method to catch a player's eye. The higher the contrast, the more likely the area is to be noticed. For example, light coming into a dark room from a doorway will attract the player's eye and that makes him a lot more likely to approach the doorway.
In general, light works really well as a method to highlight areas and paths. There's a reason that a lot of Disney theme park rides use light to mark their exits: So the visitors instinctively head toward the light and thus avoid congestion. Believe me, it works; while working on the PSP version of G-Force I had to run through a level to get to a certain point in the game to test something. To get there I had to go past a T-section in a vent shaft area.
And even though I was familiar with the level I always took the wrong turn there. Why? Because I was playing without thinking consciously, and that route was lighted better. And as soon as we had that changed I didn't end up going the wrong way.
To further expand on how important this is, there's also what's called the "squint test": you squint and look at a screenshot of your game. The brightest area is where people's eyes will most likely be drawn. There's even a squint mode feature in the Unreal Engine 3 editor that blurs your viewport to simulate such a test.
Of course, apart from light, you can also use other methods to create a contrast. You could use a unique shape that stands out to attract the player's attention. One arched doorway next to four rectangular ones will look a lot more important. You can also use color to create a contrast. A red door in an otherwise white room is going to catch the player's eye much more than a grey one will.
Similar to how the technique of contrast is informed from painting techniques, this method is too. Here, objects in the environment subtly point towards where the player has to go. The lines of the objects are oriented so that if the eye follows them it will reach the intended target. This works especially well if these lines cross the intended boundaries directly. For example, a plank lying over a chasm will give the player the hint that jumping over said chasm is the way to continue.
"Weenie" is a term taken from the Disney theme park designers. It refers to a large structure in the distance that's clearly visible to the theme park visitor. What these weenies do is give reference points for navigation within the park as whole. Additionally, they also look interesting and important so that the visitor wants to come closer and check them out. The same principle can be applied to level design.
A good example of this in use is in the game Mirror's Edge. Here, the player learns that he has to follow the red objects to get to his goal. And in one map there's a large red radio tower visible in the background. The shape is already unique as most other buildings are skyscrapers but add to this the color red which both stands out and already has some navigational meaning and this Weenie is sure to draw the player in.
Motion is an excellent way to attract attention. Based on how our human eyes and brain work, animation and movement easily catch someone's eye. This increases the effectiveness of a visual component tenfold, as compared to static clues. Whether it's sparks, moving doors or flickering lights -- they all work equally well.
However, using motion also means that player interest is a lot more conscious than with unmoving elements. This also includes scripted events such as a crane breaking down, or a door being blasted off its hinges. These events will grab the player's attention and surely make him want to investigate.
Short cutscenes can also be an excellent tool to focus a players' attention on a certain area. For example when entering a big room, the camera shows the contents of the room and ends up focusing on the closed door at the other side. This clearly tells the player that this door is important and probably the way to leave this room. Of course, this interrupts gameplay briefly -- but a cutscene can communicate complex routes quite clearly.
Adding characters to a level can help the player navigate. For example, in a shooter, the player will be used to attacking his targets and then moving past them. If he's suddenly being attacked from a balcony in a house, then this will focus his attention on that area. He might wonder how to reach this position to get a better vantage point, or to grab whatever items his enemies will drop.
Whatever the case, the player now knows that there's something up there which makes him likely to investigate. Additionally if the game features some sort of hit-direction compass then the player also gets the direction of his attackers as a clue. This makes this work even if the player's not looking at them directly when hit.
Another way to use characters is to use friendly units to show the way. If your allies are moving along the planned path and through a doorway, the player will take notice of that route and probably follow them -- though only if he notices them leave.
For this reason it might make sense to have one character leave through the doorway, and another one take up a position next to it. That way, even if the player misses the initial character leaving he can still spot the second one and come to the conclusion that the doorway is important.
Be careful with this though, as you don't want to make the player feel like he's only following his allies and not exploring himself. Also, you of course need a robust AI that doesn't get stuck or make other mistakes.