I have been reading and thinking a bit these days about the sources of startup ideas. How ideas that are worth turning into technology startups come to life. The best work I have found on this is by Daniel Gulati who actually conducted some primary research on the matter and wrote it up for the Havard Business Review. Good stuff but I don't think he has it totally right. As I have thought about this it seems there are two primary categories that serve as sources of startup ideas; those that occur naturally and those that are manufactured.
Natural or organic startup ideas just happen in the course of your life. It is almost by accident. The aha moment when you experience some pain, could not find something to fix it, and have a sudden realization that you should go out and solve the problem yourself. This is how Charles Brewer came up with the concept for MindSpring and David Cummings created Pardot (Pardot was a pivot). I believe this the best source of startup ideas. It's just kinda hard to plan an accident.
Another natural situation that occurs is when you are working at a company and discover a customer need that no one is addressing and for whatever reason your employer does not wish to address that particular market. The most obvious of these in Atlanta was ChoicePoint, which was spun out of Equifax. Endgame came out of ISS in the same fashion. ScoutMob actually got their start when EarthLink shut down its municipal WiFi network and some of the laid off employees founded SkyBlox.
The second large category of startup idea sources are those that are manufactured. The most prevalent manufactured startup idea is created of necessity. It is also known as a pivot. The original startup concept is not as successful as desired so the team has to figure out some other model. Examples abound. The aforementioned SkyBlox pivoting into ScoutMob. The Sunday Paper becoming Half Off Depot. Viture moving from a walled user generated video service to a social platform. CipherTrust was a pure message transfer agent before becoming an email security company. I could go on.
Finally there are purposely generated startup ideas. While I have an ongoing list of startup ideas that I think about, just sitting down and coming up with good startup ideas is incredibly hard. It's so hard that I am extremely hard pressed to give examples. I agree with Paul Graham (and his Ideas for Startups and How to Get Startup Ideas are most reads). The best way to get startup ideas is to look for problems. Problems where you have some domain knowledge. Problems where you know other people that might be able to validate them. Problems that you could potentially solve. While problems to solve abound not many of them meet the criteria for evaluating startup ideas.
The catch-22 around startup ideas is that in some ways purposefully generating them seems more challenging then stumbling upon one. Startup ideas are best generated if you are working on something else than trying to generate startup ideas.