Is Major Cooked? Free AI risk score + roast for your college major
Solo Era solo activities, solo maxxing, sidequests, challenge
Keepsake Project Management, Freelancers, DesignersMVP examples: 10 documented stories, and what each one tested
Well-documented MVP examples include Dropbox's three-minute demo video, Buffer's two-page landing test, Zappos photographing shoes in local stores, Airbnb's three air mattresses, and Product Hunt's email list. Each tested one risky assumption cheaply before any full product existed. The pattern is the same in all ten: a small, real thing, shown to the right people, with a result that could be measured. Below are ten, each with its source, what it tested and what you can copy.
What makes a good MVP example?
A good example shows a team choosing a cheap test for a real risk. This page keeps to stories where a reputable source describes what was built and what happened, and it leaves out the details that only circulate in blog posts. Where a source gives a number, you get the number and the source. Where it does not, the page does not guess.
If you are new to the idea, read what an MVP is first. The definition used here is Eric Ries's: the version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort. Not every story below was planned as an MVP at the time. Several were improvised, and Ries and others later used them as illustrations.
| Company | MVP type | What it tested | Source |
|---|---|---|---|
| Dropbox | Demo video | Would early adopters want it enough to join a waiting list | TechCrunch, 2011 |
| Buffer | Landing page, then pricing step | Interest, then willingness to pay | Buffer blog, 2011 |
| Pebble | Crowdfunding pre-sale | Would strangers pay before the product existed | CNN, 2012 |
| Zappos | Concierge | Would people buy shoes online | Fortune, 2012 |
| Airbnb | Real service, simple site | Would strangers pay to sleep in someone's home | Fortune, 2011 and 2012 |
| Groupon | Side project on existing software | Would group buying work with one deal a day | TechCrunch, 2011 |
| Product Hunt | Email list | Would people read a daily list of new products | TechCrunch, 2014 |
| Craigslist | Email list | Would people read local listings by email | NPR, 2020 |
| Narrow audience | Would one university sign up | Harvard Crimson, 2004 | |
| Single feature | Would people use a photo-only app | TechCrunch, 2010 |
Which MVP examples tested demand before the product existed?
The cheapest MVP is one where nothing is built yet. These three put an offer in front of people and counted the response.
01 Dropbox: a three-minute demo video
Drew Houston recorded a screen demo of a file-sync product that was still hard to build, aimed it at technical early adopters and watched the waiting list grow.
- MVP type
- Explainer video
- What it tested
- Whether early adopters wanted a product that was still technically risky to build
- Source
- Eric Ries in TechCrunch, October 2011, quoting Drew Houston
- Hard part
- Making the demo convincing without a finished product
Ries writes that the product faced significant technical hurdles and needed a highly reliable online service. To avoid years of development ending in a product nobody wanted, Houston made a video: a simple three-minute demonstration, narrated by Houston over his own screen, targeted at technology early adopters.
Houston recalled that the video drove hundreds of thousands of people to the website, and that the beta waiting list went from 5,000 people to 75,000 overnight. The video included about a dozen jokes tailored to the audience of the news site Digg, and within 24 hours it had more than 10,000 Diggs.
What to copy: show the product working as you intend it to, to the community that will care most, and count sign-ups. What not to copy: the numbers. The video worked because the product solved a problem that audience already had and the jokes were written for them.
02 Buffer: two pages and a pricing step
Joel Gascoigne tested a tweet-queuing idea with a two-page site, added a pricing page to test willingness to pay, and built the product only after both showed interest.
- MVP type
- Landing page, then a pricing step
- What it tested
- First interest, then which paid plan people would click
- Source
- Joel Gascoigne, Buffer blog, February 2011
- Time to first version
- Seven weeks of evenings and weekends, per the post
Gascoigne's idea was a way to queue tweets without scheduling each one individually. His first test was a two-page site, which he shared by tweeting the link and asking for opinions. A few people left their email and sent feedback, which he counted as his first validated learning.
To test payment, he added a page between the two showing pricing. People now had to click a plan before giving their email, which tested the price and, in his words, demand further, because it was one extra click. Some still clicked through.
He then built the first working version over seven weeks and launched on 30 November. He writes that he had his first paying customer within four days of launching the rough product.
What to copy: the order of the tests. Interest first, price second, product third. A pricing page costs an afternoon, and it answers a question that a free sign-up cannot.
03 Pebble: selling a smartwatch before it existed
Pebble put its watch on Kickstarter in 2012 and let backers pay in advance, which made demand visible and funded the first production run.
- MVP type
- Pre-sale on a crowdfunding platform
- What it tested
- Whether strangers would pay for a smartwatch before it shipped
- Source
- CNN, republishing Wired, May 2012
- Hard part
- Hardware: delivery dates and production are your problem after the campaign
The report states that Pebble passed the previous Kickstarter record of USD 3.3 million and had raised about USD 10 million with days of the campaign remaining, making it the most successful Kickstarter project to that date. The campaign officially ended on 18 May 2012.
A pre-sale tests the hardest thing to learn from talking: will people hand over money? Software teams can use the same idea with a smaller number. Offer a discounted early-access plan to a waiting list and see who pays.
What to copy: asking for money before building. What not to copy: the scale. A record-setting campaign says little about what a first small pre-sale of your own will do, and a campaign needs a clear demo and an audience to see it.
Which MVP examples were run by hand?
These founders did the work themselves, or used what they already had, before building anything that scaled.
01 Zappos: photographing shoes in local stores
Nick Swinmurn put photos of a local store's shoes online and promised to buy a pair at full price from the store whenever someone ordered it.
- MVP type
- Concierge, with a store-like front end
- What it tested
- Whether people would buy shoes online at all
- Source
- Nick Swinmurn, as told to Fortune, September 2012
- Hard part
- Paying the store full price leaves little room for profit on early orders
Swinmurn failed to find a pair of boots he wanted at a mall and wondered about an online shoe store. He went to a shoe store in Sunnyvale, California, and offered to take pictures, put the shoes online, and buy them from the store at full price if people ordered. The store agreed, and he got a few orders.
That small test came before the company. He then raised USD 150,000 from friends, co-workers and others, quit his contracting job, and in 1999 kept the same model of going into stores and taking pictures. Investors told him nobody would buy shoes without trying them on, which is the assumption the early orders had begun to answer.
What to copy: fulfil the first orders manually, even at a loss, to learn whether anyone orders. What not to copy: staying manual. The point of a concierge test is to get evidence, then decide what to automate.
02 Airbnb: three air mattresses and a simple website
Brian Chesky and Joe Gebbia offered three airbeds and breakfast to design-conference attendees in their San Francisco apartment, with a website to take bookings.
- MVP type
- Real service for a handful of customers
- What it tested
- Whether strangers would pay to stay in someone's home
- Source
- Fortune, April 2011 and May 2012
- What they learned
- The first guests were not who they expected
In the fall of 2007, San Francisco hosted the Industrial Designer Society of America conference and hotel rooms were scarce. Fortune reports that the two roommates, short of rent, procured three air mattresses and built a website for what they called Airbed and Breakfast, at USD 80 a night per guest including breakfast and sightseeing advice.
They quickly found three takers, but all were over 30, including a 45-year-old father of five, not the recent graduates they had expected. Gebbia told Fortune that was when they learned unexpected things could happen. They next took the airbed idea to South by Southwest in Austin and again had three customers.
What to copy: serve a few real customers in a place where demand is concentrated, such as a conference. What not to copy: assuming your first customers match your guess. Write down who showed up.
03 Groupon: a side project on software that already existed
Andrew Mason reused the technology of his earlier site, The Point, to start a group-buying side project that offered one deal a day.
- MVP type
- Side project on an existing platform
- What it tested
- Whether group buying could be a business model
- Source
- Andrew Mason's letter in Groupon's IPO filing, reported by TechCrunch, June 2011
- Time to commit
- Started in November 2008; fully focused on it by January 2009
Mason's letter says The Point, a site for organizing group action once a tipping point of people agree to take part, launched in November 2007. A year later he started Groupon, as a side project applying The Point's technology to group buying. By January 2009 the team had fully shifted its attention to Groupon.
The letter also says that in the early days each market featured only one deal a day, a model built around the limits of a tiny community of customers and merchants. The constraint kept the product simple.
What to copy: reuse what you already have to test a different offer. What not to copy: the pivot as a plan. It happened because users of the first product pointed the team elsewhere, which only worked because the first product had users.
04 Product Hunt: a daily email list built in 20 minutes
Ryan Hoover started with an email list of new products, invited friends to contribute links and waited to see whether people wanted the daily email.
- MVP type
- Email list on an existing tool
- What it tested
- Whether people wanted a daily list of new products
- Source
- TechCrunch, June 2014, quoting Ryan Hoover
- Time to first site
- Five days, after the list, per TechCrunch
TechCrunch reports that Product Hunt started as a Linky Dink email list in November 2013. Hoover says he created it in 20 minutes and emailed a few investors and friends, asking them to submit cool products. The list got only around 200 subscribers, but the feedback was very positive.
Hoover then teamed up with Nathan Bashaw over Thanksgiving. Bashaw put together a first version in five days. A small group of early adopters tried it first, and a week later Product Hunt became a public website.
What to copy: a small, curated audience counts as evidence. About 200 readers were enough to justify a first build. What not to copy: the network. Hoover knew investors and makers who would contribute, and an unknown founder has to find that first group deliberately.
05 Craigslist: an email list to friends
Craig Newmark began with an email list to friends about events in the San Francisco Bay Area and let readers' requests decide what came next.
- MVP type
- Email list
- What it tested
- Whether people wanted a regular local digest
- Source
- NPR, February 2020
- Start
- 1995
NPR describes Craigslist as founded by Craig Newmark, who started it as an email list to friends about happenings in the San Francisco Bay Area in 1995. NPR also notes that the site makes money by charging a small number of users to post ads such as job postings and real estate listings.
This is the least engineered entry on the list, and it is useful for that reason. The first product was a message to people Newmark already knew, which cost nothing and told him whether they wanted more. The NPR piece does not claim it was planned as an MVP.
What to copy: send your first version to a list you already have, by email, and see who replies. What not to copy: the timing. The 1995 web had little competition for local listings.
Which MVP examples started narrow or cut features?
Two famous products began by removing things. One limited its audience. The other deleted most of its own features.
01 Facebook: one university, one week of coding
Mark Zuckerberg launched a profile site for Harvard students after about a week of coding, and hundreds registered in the first days.
- MVP type
- Narrow audience
- What it tested
- Whether students would sign up for an online profile directory
- Time to build
- About a week, per the article
The Crimson reports that after about a week of coding, Zuckerberg launched thefacebook.com the previous Wednesday. It combined elements of a standard house face book with profile features that let students search for others in their courses, social organizations and houses.
As of the day before the article, Zuckerberg said over 650 students had registered, and he expected 900 would have joined by the end of the day. He told the paper he could do the project better than the university and could do it in a week.
What to copy: pick a group small enough to reach completely, and where everyone already knows everyone. A closed community lets you see whether the product spreads before it is public.
02 Instagram: what remained after cutting Burbn
Kevin Systrom and Mike Krieger removed nearly everything from their check-in app Burbn except photos, comments and likes, and what was left became Instagram.
- MVP type
- Single feature, by subtraction
- What it tested
- Whether people would use a focused photo app instead of a cluttered one
- Source
- TechCrunch, November 2010, quoting Systrom's account on Quora
- Time to rebuild
- About eight weeks, per TechCrunch
TechCrunch reports that Systrom and Krieger put about a year's worth of work into Burbn, including a complete native iPhone app. Systrom wrote that it felt cluttered and overrun with features. The earlier private web app let users check in to locations, make plans, earn points for hanging out with friends and post pictures.
They cut everything except the photo, comment and like capabilities. Systrom wrote that what remained was Instagram. TechCrunch puts the rebuild at about eight weeks.
What to copy: look at what your early users actually do and remove the rest. What not to copy: the length of the detour. This pivot came after a year of work, which is the cost an earlier test would have reduced.
What do these MVP examples have in common?
None of the ten started with a full product. Each picked a risk, found the cheapest way to test it and counted something real.
| Pattern | Where it appears | What to do with it |
|---|---|---|
| The test came before the build | Dropbox, Buffer, Pebble | Put an offer in front of people before you write code |
| A human did the work first | Zappos, Airbnb, Groupon | Deliver the first results by hand and write down what you learned |
| The audience was small and reachable | Facebook, Product Hunt, Craigslist | Pick a group you can talk to individually |
| The founders reduced scope | Instagram, Groupon | Remove what users do not touch |
| Something was counted | All ten | Choose the number before you start |
How do you build an MVP of your own?
Pick the one assumption that would sink your idea, choose the cheapest format from the list above, and decide what result counts as success before you start. The steps and a fill-in card are in what is an MVP, and the process for cutting a feature list down to one workflow is in how to scope an MVP.
If you need an idea to test, browse the board. Each idea has a generated build prompt for Claude Code, Cursor or Lovable, so a single-feature version can be a day's work. Once people use it, the next question is whether they will keep doing so, which is the subject of product market fit. For what a finished business looks like, read the SaaS examples.
Frequently asked questions
What is a good example of an MVP?
Dropbox's demo video is the most cited. Drew Houston recorded a three-minute screen demonstration for technical early adopters before the product was finished, and Eric Ries reports the beta waiting list grew from 5,000 to 75,000 overnight. The video tested demand without building the hard part. Buffer's two-page test is a second clear example with a documented timeline.
Which MVP examples did not need code?
Several did. Dropbox used a video. Zappos began with photos of a local store's shoes and the founder buying pairs at full price when someone ordered. Product Hunt and Craigslist started as email lists. Airbnb's founders offered three air mattresses and a simple website. In each case the founder tested demand first and wrote software later.
Was Airbnb's first version an MVP?
It works as one, though Fortune does not describe it that way. In 2007 the founders offered three air mattresses and a website to design-conference attendees and found three takers. They tested whether strangers would pay to stay in a home, with almost no investment, and learned their first guests were older than expected.
Do MVPs always work out?
No. The ten stories here are ones that worked and were later written up. Many tests find that nobody wants the offer, which is a useful result because it costs days instead of months. Treat the examples as methods for testing cheaply, not as evidence about your odds of success.
How is an MVP different from a startup's first launch?
An MVP is built to answer a question, and it is shown to a small group you can measure. A launch is aimed at a wide audience and assumes the question is answered. Buffer's first test went to a few people from a tweet, and its public launch came seven weeks later after the demand signals were in.
Where can I find an idea to build as an MVP?
You can browse ideas on Post your build ideas, where people post app ideas they will not build. Every idea has a generated build prompt for Claude Code, Cursor or Lovable, which helps you reach a single-feature version quickly. Test demand first with a landing page or a few conversations, then build.
Created with AI assistance and edited for this site. Examples are illustrative and do not represent customer research. About the publisher.