Most startups do not die of bad engineering. They die of building for months something nobody asked for, and finding out when there is no money left to correct. An MVP exists to avoid that: it is the cheapest, fastest way to buy an answer to the question “is this worth building big?” If your MVP is not designed around a question, it is not an MVP, it is a small product.
Week 0: the question before the product
Before drawing a screen, write one sentence: “We believe [this kind of person] will [take this action] because [this problem hurts them].” Then write which number would make you say “yes, we continue” and which “no, let’s change.” For example:
- “If 20 of the 100 people who sign up complete registration and upload their first document, we continue.”
- “If fewer than 5 do, the problem does not hurt enough or we are not solving it.”
That number is the MVP’s design. Everything that does not help obtain it gets cut.
Days 1 to 3: discovery and scope lock
Three days to do three things:
- Talk to 5 to 8 people who fit the profile. Not to ask if they like the idea (they always say yes), but to understand how they solve the problem today and what it costs them.
- Find the riskiest assumption. It is almost never technical. It is “will they pay?”, “will they switch tools?”, “will they trust us with their data?”.
- Lock the scope in writing. The smallest release that tests that assumption. One main flow, not five. This is what makes a fixed price and a fixed date possible.
Days 4 to 7: design direction
Flows and interface only for the core value, plus the seed of a design system so you do not redesign everything in phase 2. If the product has AI inside, the prototype must carry real model answers, not filler text: the user judges the intelligence, not the button.
Days 8 to 26: build, live every week
Here is the change of the last two years: with AI agents writing the first draft of the code and senior engineers reviewing and owning every release, what used to take a quarter takes three weeks. Not because more hours are worked, but because the grunt work is automated and humans only do the judgment work.
The rule that protects the calendar: every Friday there is a live URL where you click the real product. Never a progress deck. If there is no URL on Friday, something is being overbuilt.
Days 27 and 28: launch and measure
The product and its analytics go live together. Analytics is not “install Google Analytics”: it is the three or four events that answer your week-0 question, instrumented from the first commit. The first week with users produces evidence, not impressions.
What to measure (and what not to)
Yes:
- Activation. What percentage of those who enter complete the action you defined as valuable.
- Return. How many come back in week 2 without you writing to them.
- Willingness to pay. Even if you are not charging yet: how many enter a card, how many ask for a price, how many accept a paid waitlist.
- Time to value. How long a new user takes to reach “oh, this is it”. If it is more than a few minutes, that is your next sprint.
Not yet:
- Total downloads or sign-ups. Vanity if nobody activates.
- Time in app. It can mean confusion, not interest.
- What they say they would do. Only what they did counts.
The three most expensive mistakes
- Building for investors instead of users. A pretty deck with a half-done MVP convinces no serious investor; an MVP with 30 active users and an honest number does.
- Adding “must-haves”. Every feature passes the same test: does it change the answer to the question? Most do not, and the data would have killed them anyway.
- Not owning the code. If the MVP lives on someone else’s platform, phase 2 starts from zero. Repository, design and infrastructure must be yours from the first commit.
What it costs
In Mexico, a functional MVP is quoted in the market between $150,000 and $350,000 MXN and takes 2 to 4 months. We do it at a fixed price: a clickable prototype in one week from $9,900 USD, a functional MVP in four weeks for $24,900, and a production-grade one in six from $49,900. The price is locked in discovery and honored.
If you have an idea, a deck or a project another team left half-built, we start with a discovery call. You leave with the scope we would lock, the tier that fits and the date your MVP is live.