── ── Startups
Minimum Viable Product
A Minimum Viable Product is not a smaller version of the final product; it is the smallest thing you can build to generate trustworthy evidence about a single key assumption. Its purpose is learning, not shipping features. A good MVP maximizes validated learning per unit of effort and may not resemble the eventual product at all.
How it works
Start from the assumption you most need to test, then ask what's the least you must build to get a real answer. Sometimes that's a landing page, a concierge service done by hand, or a video; sometimes it's a thin slice of working software. The constraint is evidentiary, not aesthetic.
Measure behavior, not praise. An MVP that everyone says they love but no one uses or pays for has failed its job. Define in advance what result would validate or kill the assumption, so the experiment can actually decide something.
When to use it
- Testing whether there's real demand before investing in a full build
- Validating that a specific feature changes user behavior
- Reducing the cost of being wrong about an uncertain bet
- Getting to real customer evidence as fast as possible
When not to use it
In mature products where users expect reliability and a bare-bones release would damage trust more than the learning is worth.
Worked example
Zappos selling shoes it didn't own
To test whether people would buy shoes online, Zappos' founder photographed shoes at local stores and posted them for sale before holding any inventory. When orders came in, he bought the shoes at retail and shipped them, losing a little money but learning what mattered. The MVP answered the real question, will people buy, without the cost of warehouses and stock.
Why it matters for founders
Founders routinely spend months building the full vision before learning the core assumption was wrong, which is the most expensive way to fail. A sharp MVP buys that knowledge for a fraction of the cost. deciqAI's agents define the smallest test that yields trustworthy evidence before acting, so you learn the answer before you fund the build.
Install this skill (free, MIT)
npx skills add deciqAI/knowledge-skillsFAQ
How is an MVP different from a prototype?
A prototype tests whether something can be built or how it feels; an MVP tests whether real customers will actually adopt or pay for it. One answers a design question, the other a market question.
Won't a barebones MVP hurt my reputation?
It can if launched broadly to the wrong audience. Run it with early adopters who tolerate roughness for early access, and frame expectations so the learning outweighs the polish you skipped.