Fail Cheaply.

It's a given that a lot of the things you will try will not work. Just remember to always fail as cheaply as possible. Don't spend years working on an idea and perfecting it before testing it. See if you can get a prototype out quickly and test it before failing. This might mean using other people's software or a competitors product 'behind the scenes' just so you can learn cheaply.