Free Lesson
Synthetic Users in UX Research: What the Evidence Says
30 min
Oct 12, 2026 9:00 AM
By continuing, you agree to Maven's Terms and Privacy Policy.
What you'll learn
What a synthetic user actually is, layer by layer
The model, the persona prompt, the data underneath, and where each layer bends the answer.
Where synthetic users fail, and the few places they don't
What 2023–2026 studies show: sycophancy, flattened diversity, and the low-stakes tasks where they're harmless.
Read a synthetic-user study critically
What to check before you trust an accuracy number: the baseline, the task, the sample, and who ran the study.
Why this topic matters
Synthetic users are being sold as cheaper, faster research, you don't have to deal with recruitment, scheduling and no shows. The real evidence says something else. Models agree with whoever is asking, flatten the diversity you are paid to find, and sound most convincing exactly where they are most wrong. Join me to figure out why synthetic users are not a right fit for product work and what you can do instead.
