The Minimum Viable Product (MVP) is one of the most important concepts in Lean Startup methodology. A well-designed MVP tests your core business hypotheses with a dramatically simplified product that saves you a huge amount of time and effort.
The Concierge MVP, and its sibling the Wizard of Oz MVP, are clever techniques that replace a complicated technical product with humans. With minimal engineering time, you’ll be able to test the key question of your business: “does anyone even want what you’re building?”
In this guide we’ll learn:
- what the Concierge MVP and Wizard of Oz MVP are
- examples of effective MVPs
- which MVP type to choose for your business
Table of Contents
Minimum Viable Product: Brief Review
First, let’s review the MVP and what it’s designed to do.
The Minimum Viable Product (MVP) is the product that is the bare minimum needed to test your most important hypotheses. No more, and no less.
The MVP is usually a HUGE simplification from the grand product you’ve envisioned when you started your business. For example, instead of even building a functional product, you can just throw up a landing page or a video (with Dropbox being a famous example) and track user signups as a gauge of interest.
In fact, you should feel slightly embarrassed when you’ve launched your MVP – if you don’t, you’ve probably over-engineered your MVP to be more complex than it needs to be.
The MVP prevents you from wasting time building a product BEFORE you have an idea of whether anyone actually wants it. You don’t want to spend months building complicated algorithms and a beautiful interface, just to discover no sane user wants your product and your company’s entire premise is misguided.
There are multiple types of MVP’s, and which one you choose depends on your particular business and product.
The Concierge MVP and Wizard of Oz MVP are clever strategies to replace complicated technical back-ends with smart, adaptable decision makers that work right out of the box: humans.
The Concierge MVP
When you envision your company’s idealized product, you might imagine a powerful web app that gives users a smooth onboarding experience, gives them smart recommendations about what to do, follows up with nice personalized emails, and automates a high level of support.
Throw that all out. Replace it with a human. That’s a Concierge MVP.
The point of the Concierge MVP is to simplify the product by replacing automated components with humans. Every customer receives the white-glove treatment, with humans front and center delivering the service.
Not only does the Concierge MVP save you a huge amount of time by skipping the product building, it also puts you directly in touch with your early customers. By being at their beck and call, you’ll start to understand their needs deeply, driving toward your fundamental question: “do people actually want what I’m offering?”
Let’s say you’re building a dating app that matched interested daters based on their answers to a set of weird questions, like “would you rather fight 100 duck-sized horses, or 1 horse-sized duck?” These are questions that OKCupid and Match.com don’t ask. Your hypothesis is that these weird questions will let you make better matches than any other service out there.
In your grand vision, you’d have powerful machine learning algorithms churning through large datasets, spitting out the very best matches with high statistical likelihood.
But you don’t have any of this right now.
Instead, you can replace almost all of the app with a human. Instead of an algorithm making matches, you manually sign up new users and email them the questionnaires. You review all submissions manually and pair people up manually based on your criteria. Even more, you give them a true concierge treatment – you repeatedly call them to make them commit to dates, you follow up to learn how they liked their date, and you find them their next match based on their feedback.
Clearly this isn’t scalable in the long run. “How in the world is this going to handle 100,000 customers?” you think.
You don’t have this problem. When you have 100,000 customers, you can worry about the scaling problem.
For now, you’re focusing entirely on proving your hypothesis – that people want your product. In this case, you want to prove that these odd questions are the secret to better dating matches.
In the process of delivering the Concierge MVP, you’ll learn a ton about your users and the value you’re offering. Maybe you’ll find that these questions are so weird that they don’t help you make good matches at all. Even more drastically, maybe you’ll find that users don’t actually want a dating app – they want a way to make new friends when they move into a new city, and these questions are a great way to do that. This will move your company in a promising new direction.
In the book Lean Startup, Eric Ries uses the anecdote of Food on the Table, a grocery shopping service that figures out what you like eating, then compiles the shopping list at specific stores that will save you the most money. Their concierge MVP involved the founder personally scrounging coupons and compiling shopping lists for their early customers.
The beauty of the Concierge MVP is that you can test your product hypotheses without building a product at all. Because humans are delivering the service manually, you can interact personally with customers and deeply understand what they’re dealing with. This will give you a strong foundation to iterate on your product offering.
In the Concierge MVP, we replaced the automated program with a human delivering the service. The human is front and center, and the user knows she’s interacting with a human.
In some cases, you want the customer to believe she’s interacting with an automated program, not a human. But you still don’t want to build a complicated technical product. The solution is the Wizard of Oz MVP.
The Wizard of Oz MVP
The Wizard of Oz MVP wraps a technical shell around a human who’s actually powering the back-end. The customer believes she’s interacting with an automated product, but in reality a human is pulling all the levers and delivering the service.
For example, let’s say you want to build a travel service where users type a complicated request articulating what they want, and your company plans the perfect trip for them. You want to be able to handle a request like: “I want to book a 5-day, 6-night trip to Paris, in June, for two people. We’re leaving from New York. I really like food and live music, and I want a short trip to Versailles. Please minimize my total travel costs.”
This is clearly pretty tough for a machine to handle. It has to parse the language to understand what’s being said, then look up possible travel options, then assemble this into a coherent travel plan.
But a human could do this very quickly. In fact, humans do this all the time to plan their own vacations.
So to set up your Wizard of Oz MVP, you have a simple web form that takes the customer request and their email. You tell the user that your algorithm needs time to crunch through all the possibilities, and that you’ll email them back within 6 hours. Behind the curtain, a human frantically researches Paris restaurants and flights.
Finally, you email the customer back with the travel plan in what looks like an automated email. You then measure the user’s response to the service – whether they end up booking through your recommendations, and what rating they give for your service.
The Wizard of Oz MVP works because, frankly, the user doesn’t care whether the back-end is powered by an algorithm or a human. She just cares that her problem is being solved – in this case, that she gets a great travel plan that will be loads of fun.
If users want your service when a human’s behind the curtains, then you can be confident that they’ll like the automated version of your product (given that quality of service doesn’t decline).
Choosing the Right MVP
The Concierge MVP and Wizard of Oz MVP are closely related ideas and have a lot of similarities. Both of them replace complicated automated products with humans. Both test your hypotheses without needing to build a complicated technical product, saving you time and letting you iterate faster.
But they differ in one critical way – whether users know they’re interacting with a human or not. In the Concierge MVP, they do. In the Wizard of Oz MVP, they don’t.
This is important because it dramatically affects the user experience. In a Concierge MVP, you have a potential confound in your experiment – interacting with a human may make users love or hate the experience more than they would without the human. In turn, this can cause problems in interpreting your experimental results.
For instance, in the dating example above, happy users might not actually love the core of your service, which is the wacky questionnaire leading to good matches. Instead, they may love the personal attention you’re giving them. When you take out the human component in a future iteration, satisfaction will plummet.
In contrast, the Wizard of Oz MVP is a more faithful representation of your product. Aside from the fact that humans will likely be slower than machines, the user experience will largely be the same. This avoids the possible confound of the Concierge MVP.
So why use the Concierge MVP at all? Because it puts you directly in touch with customers and opens up more bandwidth to learn about their needs. You’ll have more freedom in your communication, likely leading to more brainstorming of ideas to solve your customers’ problems.
Use the Concierge MVP when you’re not strongly confident about your understanding of customer problems, and you want deeper customer interaction.
Use the Wizard of Oz MVP when you want to evaluate a faithful representation of your product and hide the human from the customer.
The concept of the MVP is to move faster toward testing your hypotheses by creating a minimal product. The Concierge MVP and Wizard of Oz MVP accomplish this by replacing a complicated technical back-end with a human.
In both methods, IF you can prove that users want your service, you can build out the automated product later.
A natural hesitation people have with these MVP’s is that it doesn’t seem scalable. Once again, you don’t need to worry about scale until you actually have too many customers to handle. This is the best possible case to have, and nearly no startup suffers from this right away.
Think about it this way – if you can’t make your product work with full human attention, your scalable product wouldn’t work either.
But if your product does work, you can work on ways to convert what the human was doing into an automated, scalable technique.