I can tell you what worked for me. The longer you can conduct it as an experiment before turning it into a business the less risky it will be.

Post informational pages on as many different possible businesses as you can. Pick things that interest you. Watch your traffic very carefully using a good stats package. Which ever pages generate the most traffic/interest develop those. Research those areas as businesses--the more research you do the more likely your success.

When the web traffic justifies it, you'll have your business and you'll know how to run it.

__________________
...the Rockies may tumble, Gibralter may crumble... G & I Gershwin, 1937