MVP Types Explained: Concierge, Wizard of Oz and Single-Feature
"Minimum viable product" has come to mean a small version of the finished app. It was meant as something broader: the smallest thing you can put in front of real customers to learn whether your riskiest assumption holds. Often that is not software at all.
The concierge MVP
In a concierge MVP you deliver the service by hand, openly, to a handful of customers. Eric Ries describes a meal-planning start-up, Food on the Table, whose founders began by personally helping a single customer plan meals and shop, before automating anything. Each step they automated was one they had already proven people wanted.
Use it when you are unsure what the right solution looks like and need to learn from close contact. Watch out for the fact that customers may love the personal attention more than the product.
The Wizard of Oz MVP
Here the customer sees what looks like a working product, but behind the curtain humans do the work. The classic example is Zappos: its founder, Nick Swinmurn, reportedly tested online shoe sales in 1999 by photographing shoes in local shops and posting them online; when someone ordered, he bought the pair at the shop and shipped it. The test answered the risky question, whether people would buy shoes without trying them on, without building warehouses or supplier deals.
Use it when the experience is clear but the automation is expensive. Watch out for scale: it only works for a small number of customers, and you must be honest about data handling.
The single-feature MVP
Build one feature that solves one painful job, and nothing else. Many successful tools started this way. The discipline is in saying no: every extra feature adds cost and muddies the signal about why people use you.
The piecemeal MVP
Stitch together existing tools (forms, spreadsheets, no-code builders, payment links, email) into something that delivers the value. It looks rough, but customers care about outcomes. Our overview of validation tools lists the categories of tools people typically combine.
| Type | What you learn | Typical time to set up |
|---|---|---|
| Concierge | What the real job is; what customers value | Days |
| Wizard of Oz | Whether the experience drives behaviour | Days to weeks |
| Single-feature | Whether one core job is enough to retain users | Weeks |
| Piecemeal | Whether people pay for the outcome | Days to weeks |
How casinos test new games: the field trial
Casinos offer a neat real-world model of the staged MVP. A new table game or a new type of gaming device cannot simply be rolled out across a state's casino floors. In Nevada, for example, new games and devices generally have to pass a supervised field trial in a live casino before regulators grant final approval, and many other jurisdictions have similar trial periods. During the trial the game is played by real customers for real money, and both the operator and the regulator watch how it performs.
Game inventors have to validate twice over: the regulator needs to see that the game is fair and works as described, and the casino needs to see that players choose it over the blackjack table next to it. Many proposed side bets and table-game variants never get past this stage. It is the closest thing the gambling world has to a concierge MVP: limited, observed, real money, and ready to be pulled if it does not earn its spot. We look at the wider testing culture of the industry in casino product design as a case study in testing.
Choosing your MVP
- Name the riskiest assumption (see how to validate a startup idea).
- Pick the MVP type that tests it fastest with real customers.
- Set a pass mark: how many customers, what behaviour, by when.
- Decide what you will do if it fails, before it runs.
