Minimum viable product (MVP): meaning, types and examples

MVP stands for minimum viable product: the smallest version of a product that lets a team test a real assumption with real customers. The point is to learn, not to ship a cut-down product.

In business, the MVP meaning is practical. You build just enough to find out whether people want what you plan to sell, before you spend months on the full version. This guide covers the definition, the main MVP types and a step-by-step build process. It also covers what to measure and the mistakes that turn an MVP into a half-finished product.

Key takeaways

An MVP is an experiment designed to test your riskiest assumption, not a smaller version of the final product.

Many MVPs involve little or no code: a landing page, a video or a manual service can be enough.

Decide what you want to learn and what result would change your mind before you build anything.

Measure behavior, such as sign-ups, payments and repeat use, rather than compliments or page views.

An MVP is not a prototype, and it is not the same as a minimum lovable product.

What does MVP mean in business?

In business, an MVP is the simplest thing you can put in front of customers to test whether your idea solves a real problem. It is a learning tool first and a product second.

The word 'minimum' means you cut everything that does not help answer your question. The word 'viable' means the result must still deliver enough value that a customer will react honestly. A product that is too thin to be useful teaches you nothing.

Testing first matters because most new businesses don't last. Of US private-sector establishments that opened in March 2013, 79.6% were still operating a year later, 50.6% after five years and 34.7% after ten, according to the Bureau of Labor Statistics.

Six traits of a useful MVP

Where does the term minimum viable product come from?

The term is credited to Frank Robinson of the product development firm SyncDev, who used it around 2001. Eric Ries popularized it in his 2011 book The Lean Startup. Steve Blank's customer development work shaped the same thinking.

In paraphrase, Ries describes an MVP as the version of a new product that lets a team gather the most validated learning about customers for the least effort. You can read more in our listing for The Lean Startup and in the wider lean startup overview.

Validated learning means evidence from customer behavior, not opinion. The method rests on that idea.

What are the main types of MVP?

The main MVP types are the concierge, Wizard of Oz, landing page, explainer video, single-feature and piecemeal MVPs. Each tests demand or value with a different level of effort.

A concierge MVP delivers the service by hand to a few customers. You do the work yourself, so you learn what customers actually need before you automate anything.

A Wizard of Oz MVP looks automated to the customer, but people run it behind the scenes. The customer sees a finished service while you test whether the outcome matters.

A landing page or smoke test describes the product and asks visitors to sign up or pay. It measures interest before anything exists. An explainer video does the same job by showing the idea in action.

A single-feature MVP builds just one function that solves one problem. A piecemeal MVP stitches together existing tools and services to deliver the experience without custom engineering.

A service MVP packages a manual service before any software exists. A consultant might sell a fixed-scope audit or a guided setup. It earns revenue and keeps the team in direct contact with customers, which suits small businesses.

Match the MVP format to what you need to learn

What are some minimum viable product examples?

Two widely documented examples are Dropbox's explainer video and Zappos's early manual fulfillment. Both tested demand before building the full system.

Dropbox founder Drew Houston posted a first screencast in 2007 and a second demonstration video in March 2008 to show how file syncing would work. Ries describes how sign-ups from the 2008 video's viewers helped confirm demand before the product was ready for the public.

Zappos founder Nick Swinmurn is widely reported to have photographed shoes in local shops and listed them online. When an order arrived, he bought the pair and shipped it. That tested whether people would buy shoes online.

Here is an illustrative example. Say a founder wants to sell weekly meal kits to office workers. A first MVP could be a one-page site and a hand-packed box for ten customers. No app is needed.

MVP ideas for different kinds of business

MVPs in healthcare and other regulated settings

What goes into an MVP, and what can wait?

Include only what the user needs to reach the core value, plus a way to record what happens. Anything that adds convenience but proves nothing can wait.

Here is an illustrative case. A meal-planning MVP needs personalized weekly menus and a shopping list. Social sharing, badges and nutrition dashboards can wait for a later round.

The opposite mistake is cutting so much that users cannot judge the offer. If the core action fails, the test tells you about your build, not your idea.

Ten questions to answer before you build

How do you build an MVP step by step?

Build an MVP by naming the assumption you most need to test. Choose the cheapest experiment that tests it, and set a pass mark in advance. Then build, run and review.

Follow these steps in order:

1. Write down your idea on a one-page plan. A Lean Canvas, a format created by Ash Maurya, is a common choice.

2. Pick the riskiest assumption. This is usually that customers have the problem and will pay to solve it.

3. Talk to potential customers first. Our guide to customer discovery tools lists the kinds of software that help organize interviews.

4. Choose the lightest MVP type that tests the assumption. If a landing page can answer the question, do not write code.

5. Set a success threshold before you start. Say you expect 30 of 500 visitors to join a waiting list. Decide in advance what number counts as a pass.

6. Run the test with a small, defined group for a fixed period.

7. Review the result honestly, then persevere, change direction or stop. Ries calls a change of direction a pivot.

MVP test card: a worked example

A test card keeps an MVP honest. It fits on one page, and you fill in the first four lines before the test starts. The example below continues the meal-kit idea from above. It is illustrative: the numbers show how to fill in the card, not results from a real test.

Assumption: office workers within two miles of the kitchen will pay $60 a week for five lunches delivered on Monday.

Experiment: a one-page site with a pre-order button, shared in three office building newsletters, plus hand-packed boxes for anyone who orders. No app, no delivery software.

Measure: paid pre-orders, not sign-ups or clicks. A second measure is how many first-week customers order again in week two.

Pass mark, set in advance: at least 15 paid orders from the three newsletters in two weeks, and at least half of them ordering again. Below 5 orders, stop. Between 5 and 15, change the offer and run one more round.

Result: say the site takes 11 paid orders and 7 of those customers reorder. Demand is below the pass mark, but repeat use is strong.

Decision: run one more round with a smaller offer, such as three lunches for $36, before building anything else. The reorder rate suggests the problem is real and the price or size of the offer is the open question.

Copy the four headings (assumption, experiment, measure, pass mark) into a shared document for each test, and add the result and decision when it ends. A stack of cards like this becomes a record of what the team has learned, which is the point of an MVP.

What should you measure in an MVP?

Measure actions that show commitment, such as payments, pre-orders, repeat use and referrals. These tell you far more than likes, page views or polite feedback.

Pick one or two metrics tied to your assumption. For a smoke test, that might be the sign-up rate. For a concierge service, it might be how many customers ask for a second order.

When you have enough traffic, compare two versions of a message or offer. Our overview of A/B testing tools explains the software category that supports this.

Expect most versions to lose. In Harvard Business Review, Ron Kohavi and Stefan Thomke report that only about 10% to 20% of experiments at Google and Bing generate positive results.

Record qualitative notes too. Customer comments explain why a number moved, which helps you choose the next experiment.

Small samples go a long way for qualitative tests. Jakob Nielsen of the Nielsen Norman Group found that a first usability test with five users finds about 85% of the usability problems, and recommends many small tests over one large one.

Six signals that mean more than page views

How does an MVP fit into agile development?

In agile development, an MVP is the first release that goes to real users, and later sprints improve it based on feedback. Agile is how you build; the MVP is what you decide to build first.

Teams often confuse the two. A team can run perfect sprints and still build the wrong thing if nobody tested the core assumption. Treat the MVP as a hypothesis and use each sprint to test a piece of it.

For more on this link between product learning and delivery, see our guide to lean product development and the lean agile software hub.

What is the difference between an MVP, a prototype and an MLP?

A prototype shows how something might look or work. An MVP tests whether customers want it in the real world. A minimum lovable product (MLP) aims to delight early users as well.

A prototype is usually not released to the market. You use it to test usability or share a concept internally.

An MVP is released to real customers, even if the group is small. They behave as they would with a real product.

An MLP adds enough polish that early adopters enjoy the experience, not just tolerate it. Some teams prefer this in crowded markets, where a rough product may be dismissed. The trade-off is more effort before you learn.

Our MVP tools category covers software that helps with each of these stages.

Prototype, proof of concept, pilot or MVP?

What are the most common MVP mistakes?

After the first test: what to do next

Frequently asked questions