MVP

MVP stands for “Minimum Viable Product”. What is meant is the most stripped-down version of an idea that you can nevertheless already offer to real users. It can do little, but that little actually works. The purpose is not to make a lot of money. The purpose is to find out whether anyone even wants to use the product. A team therefore deliberately builds less than it could, in order to get an answer faster.

Why companies deliberately build less

The most expensive mistake in the software industry is a finished product that nobody needs. Two years of work, a big team, then it launches and just sits there. This is exactly what an MVP is meant to prevent. Instead of guessing for two years, you ask the market after two months.

The term comes from the “Lean Startup” movement, which the entrepreneur Eric Ries made well known around 2011. His core idea: a business idea is at first just a guess. This guess has to be tested like an experiment in physics class. The MVP is the experimental setup, real users are the measurement.

For investors, a running MVP is a strong argument. A presentation only shows what a founder imagines. An MVP with a thousand active users shows that people use the product voluntarily. That’s why the term often appears in financial news in connection with funding rounds.

What belongs in the first version and what doesn’t

At the start there is a clear assumption, for example: “People will pay to do their tax return via chat.” Then the team strips away everything that doesn’t test this assumption. No elaborate design, no user accounts, no app for every type of phone. What remains is exactly the one path through the product that matters.

Then measurement begins. How many people try it, how many come back, how many pay? The next version emerges from the numbers and the feedback. This cycle of build, measure and learn repeats until the product gains traction or the idea is buried.

A common misconception: MVP does not mean “sloppy”. The word “viable” is crucial. An app that keeps crashing does not deliver a usable result, because users leave because of the bugs, not because of the idea. A good MVP can do little, but it does that little reliably.

MVPs in AI products and in startup reporting

Many well-known services started out tiny. Dropbox initially just showed an explainer video to test whether people would sign up for automatic file synchronization. Airbnb began with a plain website and three air mattresses in an apartment. Only once demand was clear did it become a serious platform.

With AI products, this approach is especially common. A team today doesn’t have to train its own language model, but can rent an existing one via an interface. This gets a simple assistant up and running within a few weeks. Some teams even have humans do the work behind the MVP and just pretend it’s software. That sounds like cheating, but it’s a legitimate way of testing demand.

In news articles you usually encounter the term with young companies: “The startup has launched its MVP and is now seeking funding.” Translated, this means: something real exists, but not yet a finished product. A related term is prototype, which is often only shown internally. The difference is that an MVP is in use with real customers.