Research many candidate ideas.
Don't focus on the first good idea you have.
Be systematic.
Force yourself to think of 50 to 100 ideas.
Then research each for an hour or two. This means perhaps two weeks of research time.
Map each candidate.
This might include:
- mind map of topics and subtopics
- map of top blogs
- map of books
- map of people, companies, open source libraries
- keywords
- trends
- and more.
Once you have 50 to 100 candidate ideas mapped out, then create a short list.
Then perhaps run an experiment on each short list.
Or pick one that is the most interesting and exciting and dive in.
You will be spending 5+ years with the topic, so it must be interesting enough to you that you want to work on it for that long.
- You must want to help people with the topic.
- You must want to work on it for the next five years.
- You must want to work with, talk to, help the people that work on that topic.
- You must want to read books on the topic and related to the topic.
- You must want to tinker with APIs and code on the topic.
You must really think this through.
Eventually you will hire people who know about the topic so that they can take over maintenance of the business.
You must want to work with them as well.