Wondering what is an MVP in startups before you spend months building? It is the simplest version of a product that lets you learn what real customers want, with the least effort. Build to learn, not to impress.
This page sits within the wider topic of ai startups, and it covers the idea that most early-stage product decisions start from.
| Item | Value |
|---|---|
| Year Frank Robinson is credited with coining the term MVP | As far back as 2001 |
| IMVU’s original MVP, time to bring to market | Six months |
| Earlier company’s time before launching | Almost five years |
| Time spent on one feature nobody wanted | Two weeks |
| Goal of an MVP, per Eric Ries | Maximum validated learning with the least effort |
These figures come from Eric Ries’s Lean Startup post and a 2014 essay on the MVP. Sources can be updated or reworded, so check them before you rely on the figures.
What does a minimum viable product mean?
A minimum viable product is, in Eric Ries’s definition, the version of a new product that allows a team to collect the maximum amount of validated learning about customers with the least effort. The word that matters is learning. The product is a means of getting evidence, not the finish line.
The Agile Alliance glossary adds that this learning shows up as whether customers will actually buy. It also argues that watching what people do is much more reliable than asking them what they would do.
ProductPlan’s glossary puts it in plainer terms: a product with enough features to attract early-adopter customers and validate a product idea. That makes it a test of demand, built early, before a large budget is committed.
Where the term comes from
An essay on the history of the MVP credits Frank Robinson with coming up with the term as far back as 2001. His version describes a unique product that maximizes return on risk for both the vendor and the customer. Eric Ries later made the idea popular inside the Lean Startup community.
The same essay lines up the best-known definitions, and each stresses something different:
- Frank Robinson: return on risk, for the vendor and the customer.
- Eric Ries: validated learning about customers, gained with the least effort.
- Steve Blank: the smallest possible group of features that works as a stand-alone product while still solving the core problem and demonstrating the product’s value.
If you only remember one thing, remember that all of them describe a risk-reduction tool. You invest a little, watch real behavior, then invest more.
What an MVP is not
The name misleads people. Ries writes that the MVP, despite the name, is not about creating minimal products. He also admits it is somewhat annoying, because it imposes extra overhead: you have to talk to customers or read metrics to learn anything.
The Agile Alliance names the most common pitfall. Teams believe an MVP is the smallest amount of functionality they can deliver, without the added test of being enough to learn whether the business is viable.
The word “viable” does real work too. ProductPlan says the product must be viable, meaning customers can complete an entire task or project and get a high-quality experience. An interface full of half-built tools does not qualify. The essay on the term puts it bluntly: minimum does not mean crappy, it means less than theoretically possible.
There is no formula either. Ries describes how his team’s IMVU’s original MVP took six months to bring to market, which beat a previous company that spent almost five years before launching. In another case the team spent two weeks building a feature nobody wanted, and in hindsight a simple AdWords smoke test would have shown that sooner (Lean Startup Co.). Judgment decides the right size for your situation.
Forms an MVP can take
The Agile Alliance describes the MVP as an actual product that you can offer to customers and observe. That product may be no more than a landing page, or a service with an appearance of automation that is fully manual behind the scenes. The glossary entry is clear that real behavior beats stated intentions.
Not everyone agrees on how loose the term should be. The author of the history essay notes that some in the Lean Startup community label sketches, prototypes, demos and concierge tests as MVPs, but argues the label should be reserved for real products that give users something of value to keep (Necrophone, 2014). Treat the boundary as debated rather than settled.
If you are testing an AI idea, the same logic applies. Check that people want the outcome before you invest heavily in building the technology behind it.
How do you define your MVP?
ProductPlan lays out a short sequence for deciding what goes into a first version (ProductPlan glossary):
- Make sure the planned MVP aligns with your business objectives, including your resources and any revenue goals.
- Identify the specific problems you want to solve for your user persona, and choose limited functionality using user research, competitive analysis, iteration speed and cost.
- Translate that functionality into a development plan, keeping the “viable” in MVP so customers can finish a whole task.
After launch, your customers tell you whether you have hit the bar. Ries says that if people are satisfied, you may not need to build further, and if they are dissatisfied, another MVP might be needed.
That loop points toward product-market fit, and the guide on what is product market fit explains how to recognise it. The essay on the term adds that an MVP aims to prove that fit rather than a fully scalable business model. Pricing and revenue design come afterwards, and the guide to ai startup business models covers that stage.
Examples: Airbnb and Foursquare
ProductPlan gives two familiar cases. The founders of Airbnb used their own apartment to validate the idea of online short-term, peer-to-peer rental housing. They built a minimalist website with photos and details, and found several paying guests almost immediately.
Foursquare began as a one-feature MVP offering only check-ins and gamification rewards. Its team then added recommendations, city guides and other features once an eager, growing user base had validated the idea (ProductPlan).
Frequently asked questions
Is an MVP the same as a prototype?
Show answer
Not necessarily. Some people in the Lean Startup community call prototypes, demos and sketches MVPs, but the author of one widely read essay argues the term should be reserved for real products that users can keep. Agree on a definition with your team before you start.
Do you always need an MVP?
Show answer
No. Ries says that if your goal is simply to scratch a clear itch or build something for a quick flip, you really do not need one. The MVP pays off when you must learn something from your first version.
How long should an MVP take to build?
Show answer
There is no fixed answer. Ries reports six months for one original MVP and two weeks, which was too long, for a feature nobody wanted. Aim for the shortest build that still lets you learn.
Is an MVP only for software?
Show answer
No. The history essay argues the concept works across digital products, services and hardware. What changes is how quickly a team can run the build, test and learn loop.
