# Wizard of Oz Prototype

A prototype whose apparently automated responses are produced partly or entirely by a person behind the scenes during a moderated study.

- Type: method
- Domain: Research & behavior
- Level: Intermediate
- Last reviewed: 2026-09-25
- Original: [Soroush’s Product Wiki](<https://wiki.soroush.uk/topic/wizard-of-oz-prototype/>)

## What it means

A Wizard of Oz prototype lets a participant interact with an interface or service that appears to respond autonomously while a hidden operator generates or selects responses. It tests interaction, expectations, and system behavior before expensive dynamic technology is implemented.

## Why it matters

Static screens cannot represent every conversational, personalized, intelligent, or real-time interaction. Human simulation lets a team explore these experiences and uncover requirements while the underlying system remains uncertain or costly.

## When to use it

Use it for dynamic concepts such as assistants, recommendations, data lookups, or novel input when realistic responses are essential to the research question. Use a simpler prototype when fixed states are sufficient, and review deception and safety risks before the session.

## How it works

1. Define the interactions and risks that require realistic dynamic responses.
2. Design the visible prototype, operator controls, response rules, boundaries, and failure states.
3. Rehearse timing and consistency with a facilitator, operator, and realistic tasks.
4. Obtain appropriate consent, run the moderated session, and record both participant and operator actions.
5. Debrief any material deception, analyse expectations and breakdowns, and separate prototype performance from technical feasibility.

## A practical example

A support team tests an AI troubleshooting assistant while a specialist selects responses from a controlled library. Sessions reveal that users expect evidence and escalation controls before engineering commits to a model architecture.

## Where it breaks

A skilled human can provide context, judgment, speed, or accuracy the eventual system cannot match. Operator inconsistency affects results, technical feasibility remains unresolved, and concealed human control raises ethical concerns requiring review and debriefing.

## Common mistakes

- Simulating capabilities that the team has no plausible way to deliver.
- Failing to script operator boundaries and producing inconsistent experiences.
- Treating positive interaction with a human-quality simulation as proof an automated system will work.

## Sources

1. [The Wizard of Oz Method in UX](<https://www.nngroup.com/articles/wizard-of-oz/>) — Nielsen Norman Group · Professional guidance
2. [UX Research Methods: Glossary](<https://www.nngroup.com/articles/research-methods-glossary/>) — Nielsen Norman Group · Professional reference

## Related topics

- [Prototype Fidelity](<https://wiki.soroush.uk/topic/prototype-fidelity/>)
- [Concept Testing](<https://wiki.soroush.uk/topic/concept-testing/>)
- [Pretotyping](<https://wiki.soroush.uk/topic/pretotyping/>)
