Hypothesis Refiner Blueprint
Learned of this from a friend and have been fine tuning it since.
The idea here is to fine tune create hypothesis based on 3 principles - Behaviour, User Outcome, Business Outcome.
Use whichever models you prefer but I've found that Claude and GPT5 provides a better accuracy in the end result.
Step 1
Enter the following prompt
Create a set of assumptions based on the provided information:
Behaviour assumption:
// input here: specific behaviors that, when changed, will impact user outcome
User outcome assumption:
// short or long term desires that a user hopes to satisfy through usage
Business outcome assumption:
// value created for a company from impacting user outcomes>
Then present each assumption as a short paragraph without bullet points
Step 2
Enter the following prompt
Using the these assumptions, create hypotheses that adhere to the following criteria for robustness:
Testable: Propose a relationship between variables with a method for empirical verification.
Falsifiable: Allow for conclusive determination as true or false following research.
Specific: Detailed and focused, based on in-depth understanding of context and variables.
Contextual: Grounded in existing knowledge and relevant tot he specific feature or situation.
Measurable: Quantifiable or allowing assessment of the feature's impact direction.
Then finally present the final output in this format:
Behavioural hypothesis
User outcome hypothesis
Business outcome hypothesis
Topic: Design, Discovery, UX Research
4 April2026