An MVP (Minimum Viable Product) is the simplest version of a product that allows you to launch it to market, collect real user feedback, and validate a business hypothesis with the minimum effort possible.
The concept was popularized by Eric Ries in his book Lean Startup and has become one of the pillars of modern product culture. The idea is counterintuitive for many entrepreneurs: instead of building the perfect product and launching it with fanfare, you build the minimum you need to learn whether anyone wants what you're offering.
An MVP is not an incomplete product or a low-quality version. It's a learning tool designed with intention. Every feature included (and excluded) responds to a hypothesis that needs to be validated.
90% of startups fail. The main cause isn't lack of funding or competition: it's building something nobody needs. An MVP lets you discover this in weeks, not months or years.
Building a complete product can take 6-12 months and cost tens of thousands of dollars. An MVP can be ready in 2-4 weeks at a fraction of the budget. If the hypothesis fails, you've lost weeks instead of months.
Surveys and focus groups tell you what people say they would do. An MVP shows you what people actually do. There's no substitute for the real behavior of real users with real money.
Investors prefer data over presentations. An MVP with real traction (active users, conversions, retention) is infinitely more convincing than a pitch deck with projections.
You create a web page that describes your product as if it already existed. You include a buy or sign-up button. You measure how many people click. If there's enough interest, you build it. If not, you pivot.
Real example: Buffer started as a landing page describing the tool with a "plans and pricing" button. If you clicked, you saw a message: "We're not ready yet. Leave us your email." In 7 days they had 120 sign-ups.
You create a video that demonstrates how your product will work. You don't need to build it: the video is enough to measure interest.
Real example: Dropbox created a 3-minute video showing how file synchronization would work. Overnight, their waitlist went from 5,000 to 75,000 people. Without having written a single line of code for the final product.
You offer the service manually, simulating the automation that the final product will have. You do the work that the software will eventually do.
Real example: Food on the Table started with its founder personally going to the supermarket with their first customer, planning her weekly menus by hand. Only when the model was validated did they start building the app.
Similar to concierge, but the user doesn't know the process is manual. From the outside it looks like an automated product, but behind the scenes there are people executing the tasks.
Real example: Zappos (now part of Amazon). Founder Nick Swinmurn had no shoe inventory. He photographed shoes at local stores, published them online, and when someone bought, he went to the store, purchased them at retail price, and shipped them. He validated the demand for shoes online without investing in inventory.
You build a single feature, the one you consider most critical, and launch it as a standalone product. If it solves a real problem, you iterate. If not, you pivot.
Real example: Airbnb started as "Air Bed and Breakfast": three air mattresses in a San Francisco apartment during a conference when all hotels were full. No app, no payment system, no professional photos.
What assumption, if proven false, would invalidate your business? That's the hypothesis your MVP should test.
Example: "Dog owners would pay $30/month for personalized food delivered to their door."
Before building anything, define what result you'll consider validation. Be specific: "At least 3% of landing page visitors sign up" is a metric. "People are interested" is not.
Not all MVPs are the same. Choose the format that lets you test your hypothesis with minimum effort:
| Hypothesis to Test | Recommended MVP Type | Estimated Time |
|---|---|---|
| Is there demand for my product? | Landing page or video | 1-3 days |
| Would users pay for this? | Landing with pricing + checkout | 3-5 days |
| Does the service solve the problem? | Concierge or Wizard of Oz | 1-2 weeks |
| Does the core feature work? | Single-feature MVP | 2-4 weeks |
Resisting the temptation to add features is the hardest part of creating an MVP. Every extra feature is time you're not spending learning. Ask yourself: "Can I validate my hypothesis without this?" If the answer is yes, don't include it.
Put your MVP in front of real users as soon as possible. Measure the metrics you defined in step 2. Don't get distracted by vanity metrics (visits, likes). Focus on the metrics that tell you whether your hypothesis is correct (sign-ups, purchases, retention).
With data in hand, make a decision:
Building too much: the most frequent mistake. If your MVP takes more than a month, you're probably building a product, not an MVP.
Not defining metrics beforehand: without clear success/failure metrics, any result can be interpreted as positive. Define your criteria before launching.
Asking instead of measuring: "Would you buy this?" is not validation. "I put a buy button and 5% of visitors purchased" is. Real behavior trumps declared intent.
Ignoring qualitative feedback: quantitative data tells you what's happening. Conversations with users tell you why. Combine both.
Launching to the wrong audience: your MVP should reach your target audience, not your friends and family. Feedback from people who aren't your target validates nothing.
No. An MVP can be a landing page, a video, a manual service, a paper prototype, or even a conversation with 10 potential customers. What matters is that it lets you learn whether there's real demand.
A prototype is a visual representation of the product (mockups, wireframes, models). An MVP is a functional product put in the hands of real users to validate a business hypothesis. The prototype shows what it will be like; the MVP shows whether it should exist.
It depends on the type. A landing page MVP can cost $0 (with tools like Carrd or Webflow). A single-feature MVP for an app can cost between $5,000 and $15,000. The general rule: the cost should be proportional to the risk you're eliminating.
Yes. Many companies use MVPs to validate new features or business lines. In CRO, every A/B test is essentially an MVP: you test a variant with a subset of users before implementing it for everyone.
At Boost, we apply the MVP mindset to every CRO optimization program: we test hypotheses with real data before implementing permanent changes. If you want to know what opportunities your website has, start with Scan&Boost.
— Adrià Vidal, Boost
Learn what the Aha Moment is, how to find it with data, real examples (Slack, Facebook, Dropbox), and how to use it to improve retention and onboarding.
Learn what Customer Success is, essential metrics (NRR, churn, NPS, health score), and how to implement a program that retains and grows your customers.
Learn about Jobs to Be Done, how to conduct JTBD interviews, the difference with buyer personas, and practical examples to improve your product and...
