The short version
UX research is how teams stop guessing what users want and start watching what they do. A handful of good sessions can kill a bad idea before engineering builds it, which makes research the cheapest insurance in product work.
What it actually is
UX research splits into two questions: what do people need (generative interviews, field studies) and does this design work (usability tests, task analysis). Different questions, different methods; most confusion comes from using one to answer the other.
Why small samples work
Five users finding the same broken flow is signal, not anecdote. Usability problems are structural, so they repeat fast; the classic finding is that a handful of sessions surfaces most of the serious issues, which is why speed beats scale for design questions.
How to think about it
Watch behavior, discount opinion. People are honest about their problems and unreliable about solutions, so ask about the last time they did the thing, not whether they would use your feature. The past tense is where the truth lives.
