# Hypothesis-Driven Design

A practice that expresses product ideas as falsifiable beliefs about users, changes, and outcomes, then uses evidence to guide design decisions.

- Type: practice
- Domain: Strategy & discovery
- Level: Intermediate
- Last reviewed: 2026-09-21
- Original: [Soroush’s Product Wiki](<https://wiki.soroush.uk/topic/hypothesis-driven-design/>)

## What it means

Hypothesis-driven design treats proposed solutions as uncertain claims rather than requirements. A useful hypothesis identifies the audience, intended change, expected outcome, and evidence that would increase or decrease confidence, allowing a team to learn before scaling implementation.

## Why it matters

Feature language hides uncertainty and encourages teams to judge success by shipping. Hypotheses make assumptions discussable, connect design choices to outcomes, and support tests that can change a decision. They also help teams distinguish learning goals from delivery tasks.

## When to use it

Use it when exploring a new experience, prioritizing competing product bets, or working where customer behavior and business outcomes are uncertain. It is less valuable for mandatory compliance changes whose implementation is known, though usability assumptions may still be tested.

## How it works

1. Start with the customer or business problem and the decision to be made.
2. State the target audience, proposed change, expected behavior, and outcome.
3. List the assumptions that must hold for the hypothesis to succeed.
4. Prioritize assumptions by risk and select evidence thresholds in advance.
5. Choose the smallest ethical test that can produce decision-relevant evidence.
6. Update confidence and decide to continue, revise, combine, or abandon the idea.

## A practical example

A music app hypothesizes that new listeners who can preview a personalized mix before registering will be more likely to complete signup because the value becomes concrete. The team prototypes the preview, measures qualified signup behavior, and checks whether any lift harms later retention.

## Where it breaks

A precisely written hypothesis can still rest on a weak problem frame or misleading metric. Short experiments may optimize local behavior while damaging trust or long-term outcomes. Statistical tests are not always appropriate, and failure to define thresholds invites post-hoc storytelling.

## Common mistakes

- Writing “users will like this feature” with no observable outcome.
- Choosing metrics after seeing results.
- Testing a minor interface detail while the value proposition remains uncertain.

## Visualization

[View the visualization on the website](<https://wiki.soroush.uk/topic/hypothesis-driven-design/#visualization-hypothesis-driven-design>)

## Sources

1. [The Hypothesis Prioritization Canvas](<https://jeffgothelf.com/blog/the-hypothesis-prioritization-canvas/>) — Jeff Gothelf · Creator-authored method
2. [How I Break Down Hypotheses to Make Them Easier to Test](<https://jeffgothelf.com/blog/how-i-break-down-hypotheses-to-make-them-easier-to-test/>) — Jeff Gothelf · Creator-authored guide

## Related topics

- [Lean UX](<https://wiki.soroush.uk/topic/lean-ux/>)
- [Assumption Mapping](<https://wiki.soroush.uk/topic/assumption-mapping/>)
- [A/B Testing](<https://wiki.soroush.uk/topic/a-b-testing/>)
