I find that people who use consumer research tend to fall into one of two camps. Those who look to the consumer for answers, and those who look to the consumer for problems.
People who look to the consumer for answers will spend a lot of time setting up research so that it will yield explicit recommendations for what your offering should be. I find that this can be useful if you are looking to incrementally improve your existing offering. Consumers can be reasonably explicit in telling you what they like or don't like about your current offering, and many can offer suggestions for improvements. Does it mean that you should take their suggestions literally? Not usually, but it is fairly easy to understand where they are coming from. The potential pitfall is that you could end up missing an opportunity to do something radically different, because the focus on what the consumer wants, rather than why they want it.
People who look to the consumer for problems will spend a lot of time trying to discern the motivations and values of the consumer that are driving their decision-making process. I find that this is useful if you are looking for opportunities for disruptive innovation. In this case, the consumer cannot explicitely tell you what to do, but in understanding how they make decisions, you can develop criteria for the success of a new offering. This focus can also help you to develop incremental improvements as well, because the way they make decisions doesn't change. The potential pitfall is that many people have a difficult time translating the decision process into a successful offering. It requires that the people doing the research can balance and connect linear and non-linear thinking, and they are more difficult to find.
I'm curious to know how others use consumer research, because I feel that there is a disconnect between what great research can yield, and what most companies are getting from it. Regardless of how you use research, it is a tool to help you to make decisions. If it is a stand-alone discipline it is probably too isolated to help to define either problems or solutions. What practices have you seen, and how well do they work?