Wizard of Oz

A research method where a human secretly powers responses behind a seemingly automated interface, letting teams test an idea before building the real system.

Why it matters

Why build the AI when a human behind the curtain can fake it for a week? Wizard of Oz testing answers the expensive question — would people even want this, and how would they talk to it — before a line of the hard system exists. The classic failure it prevents: six months building the automation, then discovering users phrase requests in ways nobody predicted. One ethical seam: participants are being deceived mid-study, so the debrief afterward is not optional.

Related terms

Research
Contextual Inquiry
Minimum Viable Product (MVP)
Rapid Prototyping
Tree Testing

Go deeper in Learn UX

Choosing the right Research method UX Research

Ask your assistant

Define "Wizard of Oz" using uxspot.

Works in Claude, Cursor or any MCP client — connect the uxspot MCP.

Think it stuck? A 20-second round of Spot Check will tell you.