In my article, "How to Guarantee Product Failure", I mention that a common mistake in requirements gathering is to not ask, "Why?" Prospective and existing customers often make "feature requests" and suggestions about what your product should do. If you're an executive in the company, you might conclude that you should take their advice if enough of them clamor for the same feature.
A product manager's job is to probe further into the customers' underlying concerns. Features are not requirements; they are solutions. Whenever a customer requests a feature, ask which problem this feature would help them solve. After the customer tells you a problem, ask why it's a problem. Continue asking why until you fully understand the root problem and its implications.
But when do you stop asking, "Why?" How do you know when you've fully understood the root problem? One guideline is to stop when you've reached the customer's core needs (e.g. health, sex, safety, social acceptance). Your product may address one or more problems that block satisfaction of them, but it will never fully address the core need. You know you've probed deeply enough when you reach the core.