Can you trust synthetic users?

"It really felt like you were going through a survey and reading real comments." Someone said that at Product Build Day Stockholm, after their team tested a prototype with synthetic users. More than 100 people ran synthetic tests in my rooms this year. So can you trust synthetic users? There are two questions hiding in that one: can people use it, and do people want it? Synthetic users answer the first one at light speed, and that makes your time with real people worth more. Only real people can answer the second. What you'll hear: The two camps: "fake research" against "the obvious next step" Why 100% task success is a usability result and not a market signal What synthetic users are best at for now, and what they can't do Will synthetic users replace testing with real people? When a mature product needs something else, and why someone has to read the findings with judgment Try this week: test a website you often struggle with using synthetic users. Before you look at the results, write down where you expect them to get stuck. I'm Marcin Klimaszewski and I run ⁠Product Build Day⁠. Send me your question on ⁠LinkedIn⁠, and it might become an episode. Sources: User Interviews: State of User Research 2025 (10 September 2025) uhura: AI personas and synthetic users: fast research or risky illusion? (10 September 2025) Nielsen Norman Group: Synthetic users: if, when, and how to use AI-generated "research" (21 June 2024) MeasuringU: A review of experiments with synthetic users (14 April 2026) Sauer, Braun, Kostic and Krauß: Humans vs. AI: who's the better usability evaluator? (HCI International 2026) Product Build Day September tour: Amsterdam, Copenhagen, Stockholm and Berlin