A Riskiest Assumption Test starts from assumptions rather than a preferred MVP. Priority combines impact if wrong with current evidence. Thresholds, minimum data, and decision rules are set before testing; an interview count alone is not a validity criterion.

1. 1

   Assumptions

   Break down critical assumptions

   

   Goal

   Decompose the idea into discrete desirability, viability, feasibility, and adaptability assumptions.

   Inputs for this step

   • Product or business model
   • Audience and problem understanding
   • Technical and operational conditions

   Actions

   • Ask what must be true for the idea to work.
   • Write assumptions as concrete, falsifiable claims.
   • Link each claim to current evidence.

   Output

   Assumption map

   Assumption · risk dimension · evidence source

   Quality check

   Is each claim independently falsifiable?

   Pitfall

   Broad beliefs such as 'the market wants this' bundle several unknown mechanisms. Split by segment, behaviour, price, channel, and feasibility.

   Timing: 10–15 minutes

2. 2

   Prioritise

   Select the riskiest assumption

   

   Goal

   Assess impact if wrong and current evidence separately, then justify one priority.

   Inputs for this step

   • Assumption map
   • Decision and investment size
   • Evidence sources

   Actions

   • Assess impact if wrong against the concrete decision.
   • Assess evidence by source, recency, and behavioural proximity.
   • Select high-impact, weak-evidence assumptions first and record disagreement.

   Output

   Prioritised assumption

   Impact · evidence strength · rationale

   Quality check

   Are uncertainty and importance assessed separately?

   Pitfall

   The team selects the favourite idea or easiest assumption to test. Score independently, reveal evidence, and resolve differences against the investment decision.

   Timing: 10–15 minutes

3. 3

   Test

   Design the smallest decision-capable experiment

   

   Goal

   Set evidence source, metric, threshold, and minimum data before testing.

   Inputs for this step

   • Prioritised assumption
   • Available test channels
   • Time, cost, and ethical boundaries

   Actions

   • Choose an observable outcome that directly supports or weakens the assumption.
   • Define population, procedure, metric, and threshold before launch.
   • Set stop rule and decisions for pass, fail, and inconclusive.

   Output

   RAT protocol

   Test · population · metric · threshold · decision

   Quality check

   Would a failed result change the next investment?

   Pitfall

   A fixed interview count is treated as success although statements do not test behaviour or a pre-set threshold. Define success through observable outcomes and a decision; size data to uncertainty and test type.

   Timing: 15–20 minutes

4. 4

   Decide

   Evaluate evidence and reorder risk

   

   Goal

   Apply the pre-set rule, update evidence strength, and determine the next risk-reduction step.

   Inputs for this step

   • Test data and deviation log
   • Pre-set rule
   • Assumption map

   Actions

   • Check data quality and protocol fidelity.
   • Apply the threshold and update evidence strength with rationale.
   • Decide persevere, adapt, retest, or stop, then choose the next assumption.

   Output

   Decision and new ranking

   Result · evidence strength · decision · next assumption

   Quality check

   Does an inconclusive test remain inconclusive?

   Pitfall

   The threshold moves after the test to preserve the prior belief. Keep exploratory findings separate and create a new test version for new hypotheses.

   Timing: 15–30 minutes analysis