Most founders I talk to arrive with a list. Sometimes it’s twelve features, sometimes it’s forty. Almost every item on it is a good idea.

That’s usually the problem.

An MVP isn’t a smaller version of the full product. It’s a focused version built to answer one question: do people actually want this? Everything else is negotiable.

What is an MVP, really?

A minimum viable product is the smallest version of a product that lets real people complete its core job, and tells you whether the idea is worth continuing.

That second part matters more than people expect. In CB Insights’ analysis of startup post-mortems, “no market need” has consistently ranked among the top reasons startups fail, cited by more than a third of founders. A good MVP is how you find that out early and cheaply, instead of after a year of building.

What does an MVP actually need before launch?

Less than your list. But a few things can’t be skipped.

1. One core job, done end to end

Pick the single thing someone should be able to do and make that path complete. Not three half-finished paths. One that works from the first screen to the result.

When I built Sampled Planet, the whole product started from one moment: this song sounds familiar. Every early decision, from the data model to the interface, served that moment of comparing versions of the same song.

2. The states people will actually hit

The happy path is the screen you design first. Real users hit everything around it: the empty screen before they’ve added anything, loading, errors, a search with no results, a successful action that needs confirmation.

These states decide whether a first version feels unfinished or simply early. The empty state matters most, because it’s the first thing every new user sees.

3. A fast path to value

Ask how quickly a new person reaches the moment where the product makes sense. If it’s behind a long sign-up form, a tour and three settings, most people won’t get there.

With Sampled Planet, I chose participation without a sign-up wall. Voting on a song should feel as easy as pressing play.

4. Enough trust to try it

A first version doesn’t need a full brand identity. It does need clear copy, consistent buttons and type, and nothing that looks broken. If it asks for payment or personal information, it needs to explain why. People forgive missing features much faster than they forgive feeling unsure.

5. A way to learn

Decide which behavior tells you the idea works, and make sure you can see it. That might be a simple event on the core action plus an easy way for early users to talk to you. An MVP that can’t teach you anything is just a small product.

What can wait until after launch?

More than most teams are comfortable with:

  • detailed settings and preferences
  • multiple user roles and permissions
  • admin dashboards beyond the basics
  • integrations nobody has asked for yet
  • a complete design system
  • dark mode, animations and polish beyond the core flow

Waiting isn’t the same as ignoring. I wrote about this in A Few Moves Ahead: there’s a difference between thinking ahead and designing ahead. Structure the product so these things can be added later. Just don’t build them before you know people care.

Does AI change how you build an MVP?

Yes, a lot. AI-assisted building has made a working first version dramatically faster and cheaper. I designed and built Sampled Planet myself, from strategy and data model to hosting and security, with AI tools helping move the work forward.

But AI changes the cost of building, not the cost of building the wrong thing.

When a feature takes an afternoon instead of two weeks, it’s very easy to keep adding. Screens generated one at a time also drift apart: different spacing, different button styles, different words for the same action. The product starts feeling like it was assembled rather than designed.

That’s why design matters more in AI-assisted MVPs, not less. Someone still has to decide what not to build, keep the experience coherent and make the product and design decisions that the tools can’t.

How do you know it’s ready to launch?

Put it in front of five people who match your audience and ask them to complete the core job without your help.

Research from Nielsen Norman Group has long shown that testing with just five users typically uncovers around 85% of usability problems. Watch where they hesitate, fix those spots, and test again. If people can reach the core result without you explaining anything, you’re probably ready.

Quick answers

What should be included in an MVP? One complete core workflow, the empty, loading and error states around it, a fast path to value, basic trust signals and a way to measure whether people use it.

How long does it take to build an MVP? It depends on scope. AI-assisted building can shorten a first version from months to weeks, but the scope decisions made at the start matter more than the tools.

Do I need a designer for an AI-built MVP? If real customers will use it, yes. AI can produce screens quickly, but it won’t decide what to leave out or keep the experience consistent on its own.

Launch the question, not the whole product

The best first versions I’ve seen feel small and confident. They do one thing clearly and leave room for everything the founder learns next.

Your list isn’t wrong. It’s just not version one.

If you’re figuring out what your first version should include, that’s exactly the kind of product and MVP work I enjoy most.