Founders love to roast a thousand bags before checking whether anyone wanted the flavour. They picture the warehouse, the labels, the loyal queue out the door — and then they buy the beans, hire the roaster, and build the whole supply chain for a taste nobody has actually tried. The grand opening becomes the first day anyone discovers the truth, which is a spectacularly expensive moment to learn it.
The minimum viable product was invented precisely to spare you that day. It exists to test demand cheaply: ship the smallest honest version of an idea, watch how real people respond, and let their behaviour — not your conviction — decide what comes next. That is the entire lean-startup argument, the one Steve Blank made when he insisted the lean start-up changes everything by replacing the five-year plan with a search for evidence. He had a blunt statistic to back it: most startups fail not because they execute badly but because they execute, flawlessly, on something nobody wanted. Yet the instinct to over-build is stubborn. Teams disappear for eighteen months, polishing features for customers who exist only in the pitch deck, and emerge proud of a product no one asked for. The build felt like progress. It was mostly avoidance — the comforting busywork of construction standing in for the harder, scarier work of asking.
Think of it as a sample cup. A roaster does not commit a customer to a full kilogram on faith; they pour a thimble, slide it across the counter, and read the face. The sample costs almost nothing and tells you almost everything — whether the acidity lands, whether the body satisfies, whether this person will ever come back for more. The genius of the gesture is its cheapness. Pour a hundred samples and a few will go to people who wander off; you have lost a few ounces, not the harvest. A full bag committed too early does the opposite. It does not test anything. It simply becomes expensive, beautifully packaged proof that you were wrong, sitting on a shelf nobody empties — the entrepreneurial equivalent of a warehouse full of a flavour the market quietly declined.
Here is the part founders keep getting backwards: the point of an MVP isn’t to be small. Small is a side effect. The point is to be fast at learning. A scrappy landing page, a manual back end held together with spreadsheets, a single feature shipped to ten real users — these are not lesser products, they are sharper instruments. They generate a real signal while the stakes are low enough to act on it. This is why Y Combinator tells its founders to plan an MVP around the first thing that delivers any value at all, and why Blank’s own path to the minimum viable product starts with one customer and one job to be done, not a feature list.
The cheapest version that produces a genuine answer beats the gorgeous version that produces only sunk cost. A signal you can read at week three is worth more than a launch you can admire at month eighteen, because the signal still lets you change course while changing course is cheap. The whole discipline of customer development rests on this single, slightly humbling premise — that the customer knows something you don’t, and the fastest way to find out is to ask before you build, not after you’ve shipped.
None of this is an argument against ambition. Founders who skip the sample cup are not braver than the ones who pour it; they are just paying more to be told the same thing. The market does not grade on effort. It grades on whether the flavour was right, and it will tell you for free if you let it taste a little before you ask it to buy the lot. Over-building is not commitment. It is a refusal to listen at the only price you could still afford to listen at.
So pour the sample first. Read the customer’s face — the lean forward, the polite nod, the second sip that never comes — and let that reading, not your roadmap, decide how much to brew. The whole bag is real, and one day it may well be yours. But the whole bag is a reward for being right, not a way to find out. Sell the sip. Earn the batch.
