Questions, answered

Almost. Both are quick, informal tests with people who have not seen the product. Hallway testing uses whoever is nearby, typically colleagues from another team. Guerrilla testing takes the same idea outside the office, to a café or a co-working space, to get people who are less familiar with your company.

Three to five per round. Jakob Nielsen's research found that five users uncover about 85% of the usability problems in a design, and that three rounds of five users find more than one round of fifteen, because you fix what the first round found before the next one starts.

Not for basic usability. A confusing label, a hidden button or a broken step confuses everyone. You do need target users when the task requires domain knowledge, for example a clinical workflow or a trading screen, or when you want to learn what people need rather than whether they can use what you built.

Ten to fifteen minutes per person for one task: two minutes to explain, five to ten of watching, a few minutes of notes. A remote session with a recording, such as a Hallway Test run, is capped at 30 minutes so the tester can think aloud without rushing.

It can simulate one. An AI agent can click through your flow and report what looked confusing, which is useful while you build. It cannot get confused, hesitate or misread a button the way a person does, so it does not replace the real thing. See the comparison of real users and synthetic users.