The best creations are ones that solve actual problems, rather than ones that are created just to try to create something.
For example, Aaron Fisher created the One Handed Popout. He took an existing move and changed it until it was what he needed it to be and then refined it and created a practical solution.
When one creates just to create (which can be a fun exercise) it tends to result in things that are not practical for real world scenarios.
One thing I will caution you about, though - you're probably going to create something that has already been created. Everyone does it. Magic has been around for centuries and has been practiced by some of the most intelligent and creative people in history - re-creation is bound to happen.
As RealityOne said, though - the first step is building a very solid foundation from reputable sources. Not only does this help reduce how often you re-invent things, it will also give you a bountiful toolbox to use to create.
When I want to create a new piece I first think about what I want the audience to experience. Then I write up a simple script for that experience. At that point I start thinking about methods. I'll brain dump all the methods I can think of. Then I actually start trying those methods until I find something that allows me to best create the experience I am trying to create.
This is a process that usually ends up with several iterations of methodology, often scrapping a method entirely for something else until I figure out what works best.
Thank you! One of my problems is that I don't know the plot of the trick that I want to create. How do I create a plot for a card trick?