/24

Riskiest Assumption Test

A practical riskiest-assumption test for startup ideas: identify the belief that matters most, run the cheapest credible real-world test, and decide what result changes your next move.

Start here

8 min

Open +

An idea is a stack of beliefs pretending to be one thing. The fastest useful validation work is to find the belief that is both uncertain and fatal if wrong, then buy evidence about that belief before you buy more product.

Leave knowing / Leave with one falsifiable assumption and one cheap test

The rule

Do not test whether people like the idea. Test what has to be true.

Write the idea as a set of beliefs about a specific customer, painful moment, behavior, willingness to pay, channel and delivery constraint. Then ask which belief would make the rest of the idea irrelevant if it were false.

That is the riskiest assumption. It is not necessarily the hardest technical problem. A product can be easy to build and still fail because the buyer does not care enough, cannot be reached economically, or will not change what they already do.

The best early test is the one that can change your mind before the expensive work starts.

Common assumption classes

Look for the belief hiding underneath the feature list.

Problem

The painful moment happens often enough and is costly enough to matter.

Buyer

A specific person or team has both the problem and the authority or budget to solve it.

Behavior

The customer will actually change a workflow, habit or tool to get the outcome.

Price

The value is high enough that a real buyer will make a meaningful commitment.

Channel

You can reliably reach the buyer without acquisition cost swallowing the economics.

Feasibility

The promised result can be delivered with acceptable quality, cost, speed and risk.

Four moves

Turn a vague idea into a falsifiable test.

1. Write the customer and painful moment.

Name who experiences the problem, when it happens and what they do today.

Output / One specific situation instead of a market-sized statement

2. List what must be true.

Write the assumptions that connect the problem to behavior, payment, acquisition and delivery.

Output / A visible stack of beliefs

3. Pick the fatal unknown.

Choose the assumption with the weakest evidence and the highest consequence if wrong.

Output / One riskiest assumption

4. Run the cheapest credible test.

Ask for behavior or commitment using an interview about recent behavior, a concierge test, prototype, landing-page action, deposit or paid pilot. Define the threshold before you see the result.

Output / Evidence that can change the next decision

Evidence strength

Do not call every positive signal validation.

Signal	What it can tell you	What it cannot prove alone
Compliment	The idea is understandable or interesting.	Demand or willingness to pay.
Interview about recent behavior	The problem, workaround and stakes may be real.	That the proposed solution will be chosen.
Qualified action	A real prospect will spend time or change behavior.	Repeatable economics or retention.
Deposit, preorder or paid pilot	The promise is strong enough for a concrete commitment.	Long-term product value.
Repeated paid use	The product may be earning an ongoing place in the workflow.	That every customer segment or channel will behave the same way.

Before you run it

Pre-commit the decision rule.

Write the assumption in one sentence that could be false.

Choose a test that exposes real behavior rather than asking for prediction.

Name the customer segment narrowly enough to recruit actual participants.

Decide what evidence counts before the results arrive.

Set a stop, change or continue threshold in advance.

Keep the test smaller and cheaper than the build it is meant to justify.

Record objections and negative evidence instead of explaining them away.

Do not build more because the test felt productive. Build only when the evidence earns the next commitment.

Next step / Capsize

Paid / Evidence makes the decision

Find the assumption that could sink it.

Capsize turns the idea into the assumption that matters most, proposes a cheaper real-world probe, and keeps synthetic reasoning separate from actual validation.

Pressure-test my idea →See how Capsize works

Notes / Subscribe

No schedule. No content calendar. A note goes out only when there is something worth understanding.