As someone who would actually prefer to be able to
stop getting ideas, I think my best advice would be to try to become the "real person" you're looking for.
Using your example, someone who is interested in travel will not necessarily be a travel agent, but a person who uses travel websites (of providers, travel agents, or whoever) to plan their trips. More often than not, these sites have easily identifiable spots for improvement. Page layout, search functionality, mapping...the usuals. About as common is for these sites to also be cookie-cutter sites that are either webservice or some PHP (or whatever) package that they run themselves on a webhost.
This state of commodity makes it easy to design an entire site pretty much instantly that would be an order of magnitude easier to use, because eliminating any one frustration with using a commodity and you can take entire swaths of the marketplace. In travel it's going to be all about conversion, so anything that cuts down find-and-pay time is user-friendly. I don't know how attached the inventory (destination providers, cabin landlords, etc.) gets to these sites or if they'll sign up for anything that can be explained as an improvement over the status quo, but that's easy enough to research: the next time your interest in travel takes you to a travel agent, all you have to do is wait for them to say, "Gonna take a sec, the web thingy is slow," and the conversation can flow from there. "Is it better than what you used before that?"
Furthermore, I'd bet people at (e.g.) FlyerTalk are likely to describe their frustrations with site providers in some detail, which would be another "real person" vector: go to where the customers are. If you can find people who pay money to others despite having to do it via a flawed (to the customer) solution, you might be able to find out what would constitute an improvement in the market's eyes. The company/agent/provider (inventory) side of this equation is always going to be relatively dumb.
I guess I should have a second-best advice, too: Let people tell you. They will, if you're listening. Try to speak their language rather than hunting for people who can speak yours ("it's often quite hard for them to point out inefficiencies").
So anyway, as someone who gets ideas all the time, that's a little time-lapse snapshot of my thought process when loosed upon a hypothetical.