I find that I build products for myself first - just as sysadmins write scripts to automate parts of their job, I write small apps to make my own life easier.

I then use them myself, share them with friends, and do not place any real expectation of success on them. The product I am about to launch I have been using myself for almost a year now - and I feel that it is almost ready to share with the world.

If it takes off, it takes off, if not - I will still be using it myself and I have a handful or friends who still use it.

This means you can refine a product and know that it works for at least somebody - rather than hoping you will match an idea in your head to somebody out there who needs it. And everybody belongs to a community - some friends who share interests (ie. developers who use git, tech people who read blogs) and there is no better initial trial audience than your own.

It also means that there is no expectation on a single product - more of a portfolio approach which spreads out your chances.

I will likely make five or so products available to the public in the next 6 months, if I am lucky only four of them will suck.