Critical Assumption First — Name It, Falsify It, Then Decide

Direct answer

What is a critical assumption?

A critical assumption is the one claim the rest of the plan depends on. If it is false, the other risks do not matter yet. Eric Ries, in The Lean Startup (Crown Business, 2011), calls the handful of beliefs a new venture cannot survive without “leap-of-faith assumptions,” and treats validated learning as progress measured by tests of those beliefs — not by features shipped. Steve Blank’s Customer Development, published in The Four Steps to the Epiphany (2005), puts the same idea in operating form: get out of the building and treat customer, problem, and solution statements as hypotheses. This page is the field method: one sentence, three pre-written verdicts, one cheap falsifier, one decision.

Why it matters

Why testing the easy assumption wastes the month

I have watched founders spend six weeks “validating” a claim they already believed — usually “people have this problem in the abstract” — because conversations about that claim are pleasant. Meanwhile the plan still hangs on something they have not touched: that a named buyer will pay the planned price, switch a weekly workflow, or be reachable in a channel the founder can actually work. Ries’s leap-of-faith framing is useful here because it is rude. If the business dies when one belief is false, that belief is the work. Everything else is a side quest until you have a result on that one.

The loop

The Kill-First Loop: name → cheap falsify → decide

Three moves. Do them in this order. Do not skip the writing. The loop is original to this page as an operating pack; the underlying discipline is Ries (leap of faith + validated learning), Blank (hypotheses you leave the building to test), and Fitzpatrick (questions that can produce a no).

1. 1

   Name the kill claim in one sentence

   Finish the template below before you book a call or publish a page. If you cannot finish it, you do not have a critical assumption yet — you have a slogan. The sentence must name a person, a recent pain, an observable action, a timebox, and a reason they already spend time or money.

2. 2

   Pick the cheapest test that could prove it false

   If you still do not know whether the pain is real in their week, start with conversations about the last incident — not a pitch. If strangers already describe the pain and the expensive unknown is reach, a one-page smoke test is enough. If they reach and the expensive unknown is whether they will use or pay for a delivered outcome, do the job by hand for a handful of people. Do not stand up a factory to answer a sentence.

3. 3

   Decide against the lines you wrote on Monday

   Compare the week to Continue / Refine / Stop — not to a vibe, not to a compliment, not to “we learned a lot.” If the result is Refine, rewrite the sentence. If it is Stop, stop this version. If it is Continue, the next week belongs to the next kill claim, not to a feature list.

Which one

How to pick THE assumption (not a list of twelve)

Founders often produce a wall of sticky notes and then test the note they already like. Use this filter instead. The first claim that fails a later question is usually the one.

• 1.Would the rest of the plan still matter if this were false? If yes, it is not critical yet — it is a later risk.
• 2.Is the claim about a person you can find this week? “The market is huge” cannot be falsified in seven days. “Freelance designers who chased an unpaid invoice last month will pay $29 before I automate the chase” can.
• 3.Can a no show up? If every possible outcome still lets you keep building, you wrote a slogan.
• 4.Are you avoiding it because it is uncomfortable? That discomfort is often the tell. The easy assumption — “people have this problem” — is the one founders already believe.

Worked example

One week you can copy (illustrative)

The week below is illustrative — names, prices, and counts are invented so you can see the setup. It is not a conversion benchmark and not a Yibud report. The method is what you copy, not the numbers.

• Tuesday: the cheapest falsifier is not an app. It is five conversations about the last unpaid invoice, then an offer to run this week’s chase by hand for $29. No Figma. No “AI collections engine.”
• Wednesday–Friday: five strangers from a freelance Slack, not the founder’s design-school friends. After each call: did they describe a specific invoice, what they already tried, and whether they paid or booked the manual chase.
• Saturday log: two paid $29 and sent a real invoice PDF. One wanted a $15 one-off. Two said “interesting” and did not send the PDF. The compliment column is discarded.
• Sunday: that is Refine, not Continue — the $29 subscription sentence did not survive; a paid one-off chase might. Rewrite the sentence for studios or for a one-time fee. Do not start coding reminders “so the next five have a product.”

Common mistakes

What usually ruins the week

With Yibud

Use the report to name the claim — then run the week yourself

Yibud is a free, no-signup startup idea validator. Scores come from a deterministic rule engine, not from a language model guessing success. A Startup MRI report names a single critical assumption and a 7-day action plan derived from the highest-risk dimension. That is a starting sentence, not a verdict. This page is the field method after you have that sentence: write the three lines, pick the cheapest falsifier, decide on Sunday.