Most user research fails before the first interview — vague goals, leading questions, and findings nobody acts on. Here is the lightweight process I use on client projects.
Start with a decision, not a method. "Should checkout be one page or three?" beats "understand our users." The decision tells you who to recruit, what tasks to set, and when you have enough evidence to stop.
Ask about behavior, not opinions
People are unreliable narrators of their own preferences but excellent reporters of recent behavior. Ask "walk me through the last time you booked this" instead of "would you use a feature like…". Watch what they do in a prototype; treat what they say about the future with skepticism.
A good research session ends with a decision the team was afraid to make before it started.
Close the loop within 48 hours: a one-page readout with clips, the decision it supports, and the next test. Research compounds only when findings visibly change the product.