In my experience, there are two schools of thought. You might not like my answer -
1. Ask them
Take the time to bring them into your office and watch them surf and keep it open minded. Or hire an online video review company that takes videos and screengrabs of the whole thing. It's the advanced technology that a couple businesses provide at a fraction of the cost from only a couple years ago. The trouble is, people don't know what they want. They need it pried out of their minds - seduced out of their emotions. Wyatt Jenkins of Shutterstock is a master at understanding the minds of the consumer - he has written about this some (and about philosophy #2 even more) - but in my opinion, experts who can deliver results through extensive surveys like this are very rare and expensive.
2. Don't ask them (or just ask them by whether they part with their money)
Malcolm Gladwell explains it concisely I think when he discusses spaghetti sauce. Most of the times, your customers do not know what they want. The goal here is to provide your customers with many opportunities in many places to provide their feedback. And then like 37Signals says in their super business guides, just delete their feedback. Ha! Although that's kind of out of context -- the point they are making is that if the customers demand something over and over, you won't miss their demands because they will become repetitive. And then you must deliver and wow them.
I side with philosophy number two because it is more constructive with managing risk and sides well with iterative development. In my experience, focusing on key things, key functions, and doing them better than anyone else is by far the most important. The modern way to do it, is to keep the qualitative feedback from the customer out of it, and just test them by split testing your applications. Then you'll get the "money where your mouth" feedback is, because the revenue and conversions will be driving your decision making...and thus your app's evolution.
On the other side, if you are a boutique brand, and have plans for world domination, then perhaps you should take the chance on an extensive hail mary R&D project. I don't like that idea because of the #1 rule of investing - security.