Seriously, would you consider business with anyone who worked like that? Stumbling around blindly for an idea, followed by hoping that you'll generate a set of expenses, and then finally hoping that you'll be able to earn revenues.
I worry more about you than the bloke with the family to support.
It's often difficult or impossible to figure out what consumers want, other than putting lots of ideas in front of them. Usually if you ask them, their responses will be wildly different from what they actually do. (I maintained that Reddit was a failure long after I started using it on a regular basis, and did the same with LiveJournal.)
It's not a business model that works in every field - in particular, you absolutely don't want to try enterprise applications this way (really, nobody will buy from you if you try it, as you remark). But it does work.
But I disagree that you the best approach is to throw things at the wall until you see if something sticks.
1. Build a basic version of something you think people want.
2. Try it.
3. If people don't like it, modify and try again. Do that for a few times, and if it still doesn't work. Go back to step 1.