Ask most first-time founders what their MVP's core flow is, and you'll get a feature list, not a flow. Signup, onboarding, the main dashboard, notifications, a settings page, maybe a social feed — eight things, described with equal enthusiasm, none of them marked as more important than the others. That's not indecision. It's just what an idea looks like before anyone's forced it to pick one thing to be about.
The one-feature test is how you do that forcing, on paper, before anyone writes a line of code.
The test itself
Take your feature list and run each one through three questions:
- Would someone pay you for this alone, with nothing else built? Not "would they use it" — would they actually hand over money for just this.
- If you removed it, does the product stop having a reason to exist? Not "would it be worse" — everything makes a product worse to remove. Does it stop being the product.
- Can you describe what it does in one sentence, with no "and" in it? "Users upload a contract and get payment terms extracted and can export a summary" has two features hiding behind one "and." Split it.
Most of your list survives question 1 (lots of things are nice to have paid for) and fails question 2 (most things aren't load-bearing). Usually exactly one feature survives all three. That's your core flow. Everything else just became your "not in v1" list.
Why this is harder than it sounds
The test is simple. Running it honestly isn't, because every founder has at least one feature on the list they're emotionally attached to — usually the one that was the original "aha" moment for the whole idea. That feature almost never survives question 2 on its own. It's often a differentiator layered on top of the actual core flow, not the core flow itself. The social feed is why you got excited about the product. The contract upload is why someone would pay for it on day one. Those can be the same feature. They usually aren't.
What to do with what's left
Once you have the one surviving feature, that's the sentence from the scoping brief: one user, doing this one thing, start to finish. Everything that failed the test doesn't get deleted — it goes on the explicit "not in v1" list, which is a very different thing than "cut" in a founder's head even if it's the same line in the document. The social feed might be real, later, once the core flow has proven there's a product underneath it worth adding a feed to.