MVP7 min read

Types of MVP: concierge, Wizard of Oz, landing pages and when to use each

Before writing any code, ask what the cheapest route to a true answer looks like.

Six small, different glass objects arranged in a circle, one lilac piece slightly raised

What are the main types of MVP?

Six formats cover almost every early-stage test, and each one is good at answering one or two questions and poor at the rest. This is our side-by-side summary.

Six types of MVP compared (our summary)
TypeHow it worksWhat it testsEffortMain limitation
ConciergeYou deliver the service by hand, openly, to a few customersWhether the problem is real and the outcome is valuedLittle money, lots of timeDoes not scale or test the interface
Wizard of OzThe interface looks automated; a person works behind itWhether users adopt the solution as they will eventually see itMediumHuman cost per request; careful data handling
Landing page or smoke testA page describes the product and asks for an action: sign-up, booking, choosing a planInterest and message clarityLowA click is not a purchase
Explainer videoYou show how the product will work before it is readyWhether the audience gets it and wants itLow to mediumNeeds an audience you can already reach
PiecemealExisting tools stitched together: forms, spreadsheets, automations, paymentsWhether the end-to-end flow works and has valueLow to mediumBreaks beyond a few dozen users
Single-featureOne core function, properly builtRepeat usage and willingness to payMedium to highCosts more and needs solid foundations

Which types of MVP need no code?

Concierge, landing page, video and piecemeal MVPs need little or no software. They are the fastest way to learn whether the problem exists and whether your pitch lands with the right people. Y Combinator's Michael Seibel makes the point that some startups can start with nothing more than a landing page and a spreadsheet.

Concierge MVP

You do by hand what the product will one day do, and the customer knows it. In Do Things that Don't Scale, Paul Graham recalls that Viaweb built online shops on behalf of merchants who were reluctant to sign up themselves, and that doing the work manually teaches you what to automate later (Graham, 2013).

Say you want to sell weekly training-and-nutrition plans to amateur triathletes. Before building an app, write the plans yourself for ten paying athletes and send them by email. If eight are still paying after a month, you have learned more than any survey could tell you.

Landing page and smoke test

A page describes the product as if it existed and asks for a concrete action. Buffer, the social media scheduling tool, put a plans-and-pricing page between its landing page and the sign-up form; clicks on paid plans signalled willingness to pay rather than idle curiosity (Buffer). How to build one that gives trustworthy answers is covered in landing pages to validate a startup idea.

4 days

from Buffer's public launch to its first paying customer, after a landing-page and pricing-page test had shown people clicking on paid plansSource: Buffer, Idea to Paying Customers in 7 Weeks: How We Did It

Explainer video

Dropbox is the textbook case. As Eric Ries recounted in TechCrunch, Drew Houston recorded a roughly three-minute screencast showing how file syncing would work, before the product was open to everyone. According to Houston, the beta waiting list jumped from 5,000 to 75,000 people overnight.

It worked partly because it reached the right crowd, Digg's technical community, with in-jokes aimed squarely at them. Without distribution, an explainer video is just a file on a server.

Piecemeal MVP

You combine existing tools, such as a form builder, a spreadsheet, an automation service and a payment link, to deliver the full service. The customer sees something that works; you see exactly where the flow sticks. It holds up while volumes are small, which is all you need right now.

What is a Wizard of Oz MVP, and when should you use one?

It is an MVP where users interact with what looks like an automated system while a person quietly does the work behind it. The term comes from interface research: Nielsen Norman Group traces it to Jeff Kelley, who coined it in 1983 at Johns Hopkins while studying natural-language interfaces.

Graham's essay has a startup example too: early on, Stripe signed merchants up for traditional merchant accounts by hand, behind the scenes, so that accounts looked instant to users. Today it is the most honest way to test an AI product before you build the automation: a person does what the model will do, and you measure whether the result actually matters to people.

When is a single-feature MVP with real code worth it?

When you already have evidence that the problem exists and the open question is repeat usage: will people come back every week without being chased? A concierge test answers that poorly, because you are the one doing the work.

A single-feature MVP builds one function well, on foundations fit for real users: secure sign-in, protected data, backups and analytics. It is the most expensive type; the cost drivers are in MVP development cost and the full method in what is an MVP.

Which type of MVP works best for B2B?

Concierge and Wizard of Oz tend to win in B2B, because customers are few, each one is valuable and a direct relationship beats traffic. Ten companies using a hand-run service and asking for a contract are far stronger evidence than a thousand newsletter sign-ups.

The next step is often a pilot with one customer, which is not the same thing as an MVP; we explain the difference in MVP vs prototype vs proof of concept.

How do you choose the right type of MVP?

Start from the question you need to close, not from the tool you like best.

Choosing an MVP type in four steps

  1. Write down the question

    Does the problem exist? Does the message land? Will they pay? Will they come back? Each type answers one or two of these well and the rest badly.

  2. Start with the cheapest format that answers it

    If the question is whether the problem exists, a concierge test is enough. If it is whether people will pay a given price, you need at least a pricing page or a pre-order.

  3. Set the bar in advance

    Write down which result counts and which does not, with a number you choose before you see the data. Without a threshold, every result looks encouraging.

  4. Move up only with evidence

    Go to a Wizard of Oz or a single-feature build once the previous rung has given you a clear answer. The wider method is in how to validate a startup idea.

Tempted to skip the ladder and build straight away with AI tools? Read how to build an MVP with AI first: the prototype arrives quickly, but the question you need to answer has not changed.

Frequently asked questions

What is the difference between a concierge MVP and a Wizard of Oz MVP?

With a concierge MVP the customer knows you are doing the work by hand. With Wizard of Oz they use an interface that looks automated and never see the person behind it. The first tests the value of the outcome; the second also tests the experience.

Is a landing page an MVP?

Yes, if it asks for an action that costs the visitor something, such as picking a plan, booking a call or pre-ordering. A page that only collects email addresses measures curiosity, which is useful but weak evidence.

Is a Wizard of Oz MVP deceptive?

Not if the service people receive is real and you handle their data as your privacy notice says. It becomes a problem if you promise capabilities that do not exist or hide who can see their data; if unsure, ask a lawyer.

Which type of MVP is cheapest?

In cash terms, landing pages and concierge tests: free or cheap tools plus your time. A concierge test eats hours, though, so cap it to a few customers and a fixed period.

Can you combine different types of MVP?

Yes, and it is common: a landing page to gauge interest, then a concierge service for the people who signed up, then a Wizard of Oz to test the experience. What matters is that each stage has its own question and its own threshold.

Sources

  1. Paul Graham, Do Things that Don't Scale (2013) (19 October 2026)
  2. Eric Ries, Dropbox's minimal viable product (TechCrunch, 2011) (19 October 2026)
  3. Nielsen Norman Group, The Wizard of Oz Method in UX (2024) (19 October 2026)
  4. Buffer, Idea to Paying Customers in 7 Weeks: How We Did It (19 October 2026)
  5. Y Combinator, Startup School Week 2 Recap: Michael Seibel, How to plan an MVP (2019) (19 October 2026)

Figures and definitions checked on the dates shown.

Building a startup?

Let's talk