[tmaly]

I ran into the same situation many times. Come up with an idea, buy a domain, do some coding, drop the idea.

You really have to stick it out and work through the idea. The best way to approach it is to do customer development first. Steve Blank was probably the first to bring this idea to the forefront.

Find out if people actually have a problem for which you are building the solution. The book Running Lean has some scripts you can mirror to do this. It will save you lots of time and actually get you pointed to an idea that has some merrit.

Derek Sivers from CDBaby had a great approach to this where he contrasted pushing onto people verse pulling an idea from people where there is already demand.