Ask questions until you know enough that you can qualify the project as taking "a long time", "not that long" and "not long at all". It's not as simple as looking at the number of elements on a screen.
Here are some questions I tend to ask clients when they approach us:
- Who is the target audience? - so we know what kind of people will be using the interface
- Who are your competitors? - so we can learn more about the problem domain and common solutions in this market or industry
- What kind of websites do you like/visit? - so we know what kind of expectations and experience the client will have
- What is the budget? - so we know what kind of UI we can build
The more questions you ask, the more you learn about the client's background and domain. That knowledge allows you to ask even more specific questions until you reach the stage where you feel comfortable enough to start forming screens in your head. At that point, it becomes gut feeling. Is the problem domain really complicated? Do users need to do a lot work? Then you might be looking at a "Complex" screen design. But watch out - you might be able to create simple screens for a complex problem as well. The best way to find out is to start iterating, testing, and creating prototypes as soon as possible.