Start with the Riskiest Assumption, Not the Full Product

Top founders do not begin an MVP by asking, “What features should we build?” They begin by asking, “What is the single most dangerous assumption in this business?” That assumption might be demand, willingness to pay, supply, retention, or delivery. The MVP exists to test that assumption as cheaply and quickly as possible. Eric Ries popularized the term minimum viable product, but the founders who ship fast interpret “minimum” aggressively: the smallest thing that can produce real evidence. Airbnb’s first version was not a global travel platform; it was air mattresses in a founder’s apartment. Zappos founder Nick Swinmurn did not build a warehouse and inventory system first; he posted photos of shoes from local stores and bought them at retail when customers ordered. Dropbox tested demand with a simple demo video before building the full sync engine. Stripe made it radically easier for developers to accept payments, but the early product focused on one painful job rather than every financial workflow. The lesson is not that founders should avoid building. The lesson is that they should refuse to build anything that does not directly reduce uncertainty. Before writing code, write down the riskiest assumption, the experiment that will test it, the metric that will decide success, and the time box. If the riskiest assumption is that people will pay, pre-sell. If it is that hosts will list, recruit ten manually. If it is that users will return, ship one core loop and watch retention. A fast MVP is not a smaller version of a polished product; it is a targeted answer to a specific question.

Use Concierge and Wizard-of-Oz MVPs Before You Automate

Some of the fastest MVPs look nothing like software. They are services delivered manually behind the scenes. A concierge MVP means the founder delivers the value by hand, while a wizard-of-Oz MVP means users interact with what appears to be an automated product, but humans do the work behind the curtain. Both approaches let founders learn without waiting for engineering. Airbnb’s founders famously went door to door in New York, photographed listings, and helped hosts write better descriptions. They did not start with a perfect host onboarding system; they started by doing the work themselves. Uber’s early operations involved manually dispatching cars and recruiting drivers city by city. Zappos did not automate inventory before proving that people would buy shoes online. Food delivery, coaching, design services, and marketplace startups have all used concierge versions to learn the real job to be done. The benefits are speed and intimacy. Founders hear complaints directly, see where users get confused, and discover edge cases that no product spec would predict. They can test pricing, messaging, and quality before writing a single line of production code. The danger is staying manual too long, so the rule is simple: automate only after the manual process reveals a repeated, high-value workflow. Use the concierge phase to learn the script, the quality bar, and the metrics that matter. Then automate the bottleneck, not the fantasy. In the early days, founder time is often the most valuable feature.

Launch to a Narrow Audience and Learn in Public

A fast MVP does not need a big launch. It needs the right small audience. Facebook started with one campus. Airbnb focused on New York before expanding to other cities. Uber began in San Francisco. Superhuman used a waitlist and onboarding calls to control growth and learn from every user. Top founders know that broad attention too early creates noise, support burden, and false signals. A narrow audience gives higher signal because users share context, can be reached directly, and are more likely to forgive rough edges. The goal is not ten thousand signups; it is ten to fifty users who use the product repeatedly and tell you the truth. Founders should recruit these users personally. Send direct messages, join niche communities, run founder-led sales, and onboard the first users yourself. Then learn in public. Share build updates, ask for feedback, publish what you are testing, and explain what you changed. This does not just attract early adopters; it forces clarity about the problem and creates a feedback loop that is faster than any survey. Ask users: What made you sign up? What almost stopped you? What would you miss if this disappeared? Watch them use the product, not just listen to their opinions. Brian Chesky has said it is better to have one hundred people who love you than a million people who sort of like you. For an MVP, that is not a slogan. It is the operating model. If the first ten users do not care, adding more users will not fix the product. Narrow the audience, deepen the relationship, and let word of mouth compound only after retention appears.

Measure Learning Speed, Not Feature Count

The fastest founders optimize for learning speed, not output volume. They know that shipping twenty features can be slower than running five focused experiments, because features create maintenance, complexity, and false confidence. The key metric is cycle time: how long it takes to form a hypothesis, build a test, put it in front of users, and decide what to do next. Reid Hoffman’s famous line—if you are not embarrassed by the first version of your product, you have launched too late—captures the bias toward action. But speed does not mean recklessness. It means choosing the smallest test that can change your mind. Track one primary metric per stage: activation, retention, referral, or revenue. Use cohort analysis instead of vanity totals. Talk to users weekly. Set kill criteria in advance so that a failed experiment becomes a fast pivot instead of a slow death. Slack began as a failed game company and pivoted when the internal communication tool showed stronger pull. Instagram grew out of Burbn, a check-in app stripped down to photos, filters, and comments. YouTube started as a dating site before becoming a video platform. In each case, the founders did not fall in love with the original plan; they fell in love with the evidence. Technical debt is acceptable in an MVP if it is documented and contained, but core risks such as payments, privacy, and security cannot be ignored. The discipline is to ship small, measure honestly, and decide quickly. An MVP is not a badge of shame. It is a compressed learning loop. Build fast enough to be wrong sooner, learn fast enough to become right sooner, and let user behavior—not internal opinion—set the roadmap.