For example, maybe you have ideas for products in half a dozen industries, but in one of those industries you happen to have a good friend who works in it (and thus has contacts there) and know the editor of one of the industry trade magazines. Those are both delivery channels into that industry. Compared to another niche where you have no contacts, it is much better to build a startup for that niche where you have some way to reach the market.
This may be already implied in your mind, but a lot of people wouldn't get it from reading your article. They'd start listing out industries and then picking one of them. I'm suggesting: list out your deliver channels first, and go from there (as a kind of step 0 in your process).
For example, Stack Overflow had two excellent delivery channels to reach programmer communities around the world (Joel and Jeff's blogs). It's fair to say that those two channels "made" stack-overflow.
Then follow the one where you best succeed into building delivery channels.
Best of both worlds?