One of the best ways to identify real problems in a market is to be a consultant and actually get paid to help solve the problem. Many great product companies started off this way (ex: Freshbooks, Hootsuite, Shopify, etc). They solved their own problem, and then build a solution so that others could use it.
If you're already a startup, and you're trying to refine and learn, then you can do more hands-on deployments for customers, where you actually go in and help them solve the problem with your software. Now, if you start to get paid by the hour to do this work, then you went to far.
Either your a service company learning and getting paid, so that one day you can spin out a product, or your a venture backeable startup and doing things that don't scale initially to learn how the customer thinks.
The best question I've ever heard to truly understand a customer is to ask "What do you do 3 minutes before, and after, you use our product". That's where the opportunity lies.