MVP – A misunderstood concept.

“Minimum viable product” (MVP) has become of the most overused terms in Silicon Valley. If you walk into any of Palo Alto’s many coffee shops you’ll hear the term being used by wannabe entrepreneurs at nearly every table. 

But what is an MVP, exactly? Some people seem to think it’s a pretty slide deck of their startup idea; some think it’s a nice design mockup of the product; some people seem to actually be writing software code.

As Eric Ries explained in his book, The Lean Startup, “The minimum viable product (MVP) is that version of a new product a team uses to collect the maximum amount of validated learning about customers with the least effort.” Learning about customers is the important part of that sentence. That’s the purpose of an MVP.

History is littered with the dead bodies of startups that built a product no actual customers wanted. Some spent millions of their investors’ money proving that there was zero market demand for their product (I have personally done that and can report that it was painful). The value of an MVP is to avoid that sort of startup death by creating a bare-bones product first, getting it in front of real customers to see how they react, and then building out and refining the product in a way that is informed by actual customers.

Tesla launched by buying car bodies and chassis from Lotus, putting Tesla’s EV motors in them, and selling them branded as the Tesla Roadster. They made and sold five hundred of those, and from the learnings gained, they developed the Model S, the first car that was actually fully their own design.

Peloton launched by creating a picture of a prototype and making it available for pre-sales on Kickstarter. 297 people paid $1,500 each to get one of the first models, and based on the feedback from that first production run, they built a company now worth over three billion dollars. 

The concept of an MVP is really simple: get an early, imperfect version of the product out into the market so you can learn from actual customers as you refine the product offering.Yet there still seems to be confusion about what an MVP is, so here are my own thoughts. First, let me tell you (from my personal experience) what an MVP is not:

1. An MVP is not something you build to show; it’s something you build to learn from. Many people working on their MVPs think they are building a demo to impress someone. That’s nice, but the purpose is to learn actionable insights from real customers. 

2. It’s not going to get you funded. No doubt you’ve heard someone say, “As soon as our MVP is complete, we’ll show it to venture capitalists and raise lots of money!” In truth, this is backwards thinking because it implies that the purpose of an MVP is to show something. Showing that you can create a clickable prototype isn’t going to impress investors—any fifth-grader can do that. But proving to them that you have discovered key insights into the wants and needs of real-world customers by having an initial product out in the actual marketplace? In investors’ minds, that’s golden. 

3. It’s not even a thing; it’s a process. If you embrace the MVP concept, you are actually embracing an iterative process of idea generation, data collection, analysis, and learning. You’re embracing the concept of optimizing your path to product-market fit by getting early product releases out into the hands of customers so you can learn from them. Marc Andreessen has written that, if you do this right, “the market actually pulls the successful product out of the startup.” That’s what you want from your MVP. Twitter thought they were developing a group SMS platform when they released their MVP at the 2007 SXSW festival, but the market pulled out a real-time media platform that has been a billion-dollar success. 

Now I’ll tell you (again, from my own personal experience) what a good MVP is:

1. It’s Simple. When DoorDash launched, it was a simple landing page with PDF menus and a phone number that the founders answered themselves. It probably cost less than one hundred dollars to build, but the learnings they got from it allowed them to build a company that has now raised almost 190 million dollars.

2. It tests the right things. When Pandora first launched, the playlists were manually managed by humans. The question they needed to test was: Will people pay for a streaming music service? (Note that the test question was not: Can we write a software algorithm that chooses good music?) Once they successfully proved market demand with their MVP, they made the investment to write the complex software required to algorithmically choose music.

3. It’s built for rapid iteration. The whole idea of an MVP is to rapidly try out new features and see how the market responds. Rapid iteration accelerates the process of getting from your guess as to what the market wants to something that the market has proven it wants. 

Want one more example? Here’s one you can eat.

My favorite kind of MVP is food trucks. For a relatively small investment, an aspiring restaurateur can take her menu on the road, get feedback from real customers, create daily menu iterations based on that feedback, and then—after having proven market demand—she can make the investment in leasing restaurant space and building a kitchen and full dining room. That’s how a good MVP should work.

Remember my favorite Steve Blank quote: “No business plan survives first contact with actual customers.” The entire purpose of an MVP is to avoid this by getting early learnings from actual customers before you start executing a plan.

Put Some Turk in Your MVP

In the eighteenth century, a guy named Wolfgang von Kempelen created a machine that seemingly could play chess. It was a large wooden box with a chessboard and a Turkish-looking mechanical man sitting at it, ready to place chess with anyone. As a person made moves on the chessboard, the Turk would reach out its mechanical hand and make its counter-move. This amazing device was demonstrated in all the royal courts in Europe and became known as the Mechanical Turk. People were astonished and entertained by the magic of a mechanical figure that could play chess and make smart moves.

Eventually, of course, it was discovered that there was actually a human hidden inside the box who was controlling the Mechanical Turk. It wasn’t magic, after all, but everyone still loved it.

There are several examples of startups using this approach for building an MVP that looks like there’s magic going on, but the magic is actually being done manually. I mentioned previously how, when Pandora launched, it was humans manually creating the playlists, even though users thought the software was personalizing playlists on the fly. 

When Wealthfront launched their automated investing platform, the “automation” was just plain old humans sitting at desks. In both cases, they were able to get learnings from their Mechanical Turk MVPs that then informed the building of the actual automated product.

Zappos began as a website called ShoeSite.com. Founder Nick Swinmum would take pictures of shoes at stores around town and post them to the website. When orders came in he would litterally drive to the store, buy the shoes, and manually ship them to the customer. Customers thought it was a slick digital process, but it was actually just Nick, doing everything manually and learning as much as possible about customers. Based on the learnings he changed the branding to Zappos, met Tony Hsieh, raised the nessesary capital, and then actually built out the digital process and inventory that is Zappos today. 

Sometimes this is called a Wizard of Oz MVP, since it appears to be “great and powerful” but it’s actually just an average dude behind the curtain. Whether you want to call it Wizard of Oz or Mechanical Turk, many great startups have used this approach to create an MVP for a new product or service. 

Whatever kind of MVP you decide is right for your startup,  remember that the idea is to build something from which you can learn about customers and refine your product offering via an iterative process as you march your way toward Product-Market Fit.