I hate the word "tricks." To me, that reflects a mindset that takes the magic out of what we do. At a minimum, let's call them "effects" which reflects what the audience sees or "methods" which reflects what we do. I like the idea (from Larry Haas' book Transformations) of talking about "performance pieces." I just had to get that out of my system.
I develop new performance pieces based on what I want to perform. I don't develop new pieces to sell them and make money. If I'm going to put in the effort to make something unique, I want to perform it.
I think the most important thing is to have a strong basis in understanding of both the basic and advanced principles for the type of magic you are doing. Those principles are your toolbox. If the only sleights you know in card magic are a double lift and a pass, then you are going to be extremely limited in what you can do. So for card magic, I would hope you've gone through the full five volumes of Card College and some books on self-working effects (Scarne on Card Tricks, Encyclopedia of Card Tricks, Card College Light, Fulves' Self Working Books) so that you undertstand all of the available tools.
If you are creating a variation, make sure you know everything else that is out there. Research the effect using the Behr Archive (
http://www.conjuringarchive.com), MagicPedia, the Genii Forums and the Cafe and then read the versions of the effect referenced. Then decide why you need a variation? What is wrong with the effect that you are improving?
My thought process is primarily about the effect, methods and trade-offs. I start with a general idea of what I want to do or the "effect." It has to be interesting and something that appears to be magical. I then explore possible alternative methods and decide which method works best for the effect and my style. For example, I'm working on an effect where I need a selected card to vanish from the deck. I came up with several methods. Some involved gaffed decks, some involved gaffed cards, some involved sleight of hand, some involved the power of suggestion. I decided on the option that used the gaffed card because it was the cleanest for a surrounded parlor performance, I could justify using the card in the selection process and because the I was comfortable with the level of spectator management that was required. In designing any effect, each method has trade-offs -- it is a matter of whether those trade offs matter for the effect being designed or for your style.
At this point, there are a lot of things to look at to see if a method will work. Is it practical? Is it angle sensitive? Is it sleight heavy? Is there a possibility of failure? Does the handling arouse suspicion? Is the handling justified based on the construct of the effect?
After deciding on the method that works the best for the effect, I then construct the prsentation or script. For me, merely saying what I'm doing isn't what magic is about. The presentation has to be engaging and entertaining.
Once I've got an effect that is interesting, a method that is the best possible and a presentation that is entertaining, I think I've got a good presentation piece. At that point, I practice the method (i.e. technical learning of sleights) and then rehearse the presentation (sleights along with scripting as if I'm performing). Then I perform for my wife and son. Review, revise, rehearse and then it is ready to be performed for audiences. Their reactions ultimately determine whether it is good. After each performance, I critique, revise, rehearse and then get ready to perform again.