// description
Wizard of Oz prototyping creates the illusion of a functioning system by having a human behind the scenes manually perform what the technology would eventually do. The user interacts with what appears to be a complete product, but a person is doing the processing in real time. This allows teams to test whether users want and understand a concept before investing in the engineering required to build it.
// history
The name comes from the 1939 film in which a man behind a curtain operates the apparently powerful Wizard. The method has been used in computer science research since at least the 1980s (notably by Jeff Kelley at IBM) and is a standard tool in design thinking and lean startup practice. It is especially common when testing AI, voice, or recommendation features that would be expensive to build.
// example
A creator wants to test whether her community members would pay for a "personalised KDP niche recommendation" service. Instead of building an algorithm, she sets up a simple intake form. Users submit their skills and interests, and she manually sends each person a personalised niche recommendation within 24 hours — framing it as a "beta AI tool." After two weeks with 20 beta testers, 14 say they would pay £15/month for the service, and their responses reveal exactly what inputs and outputs they value most. She now has validated demand and a clear product specification before writing a single line of automation code.
// katharyne's take
This technique is perfect for validating digital product or service ideas before you build them. Want to know if people will pay for a KDP keyword research service? Do it manually for 10 people first. Want to test a new course format? Run it live on Zoom before you record a single video. The Wizard of Oz approach lets you learn from real customer behaviour without the sunk cost of a fully built product. It's how I've validated every major product idea in my business — the manual version always teaches me something the theoretical version misses.
// creative uses
- Validate a new course idea by offering a "workshop" version first: create a Kajabi or ThriveCart checkout page, set the price, send the link to your list, and run the first session live on Zoom using slides and screen sharing. Only record and package into a self-paced course if the live version sells and students complete it. You've tested demand, format, and content quality before producing 10 hours of video.
- Test a KDP niche subscription service by doing the deliverable manually for your first five subscribers: they pay via Stripe, you email them a manually-researched niche report each month. After three months you know exactly what format they value, what questions they ask, and whether retention is strong enough to justify automating. Most automation projects built before this point would have automated the wrong thing.
- Use Wizard of Oz to test a Midjourney prompt pack before you produce the full version: offer "custom Midjourney prompt consultations" at a fixed price via Calendly. Run five sessions manually, writing personalised prompts for each customer's niche. The patterns across those five sessions tell you exactly what to include in the prompt pack, and your five customers become your first reviewers.
// quick actions
- Identify your next planned digital product or service idea. Ask: can I deliver this manually for 5-10 people before building the full version? If yes, do that first. Set up a simple order form (Typeform + Stripe, or Gumroad with manual fulfilment), price it at your intended launch price, and offer it to your existing audience. Their willingness to pay at full price is the only validation that matters.
- Before building any automation (Zapier sequences, Kajabi funnels, ConvertKit workflows), run the process manually three times end-to-end. This teaches you what inputs vary, what decisions need human judgment, and what the output actually needs to look like. Automating a process you haven't run manually first is almost always the wrong automation.
- Run a "beta offer" this week for something you've been planning: one Google Form for applications, one Stripe payment link, you deliver it manually. The beta validates demand, generates testimonials, and gives you the real product brief — all before you've invested more than a few hours in infrastructure.
// prompt ideas
I have an idea for a [digital product / service / subscription] for [describe your audience]. Help me design a Wizard of Oz prototype I could run this week to test demand before building anything. What should the intake process look like, how should I frame it to customers, what should I deliver manually to simulate the finished product, and what am I trying to learn from the first 5–10 participants?
Before I build out my [course / membership / Etsy product line], I want to validate it the cheap way. Design a manual beta offer I can sell this week: write the pitch copy, suggest a tool-free fulfilment method, and give me a list of 5 questions I should ask participants after they've received the deliverable — so I learn what they actually valued versus what I assumed they would.
I want to test whether my [community / audience / email list] would pay for [describe your service idea]. Write me a one-paragraph announcement I could send to my list today that presents this as a "founding member beta" opportunity — honest about the manual/beta nature — with a clear price and a simple way to sign up. Then give me the 3 metrics I should track over the first 2 weeks to know whether to build the full version.