hypothesis

Files (3)

• SKILL.md
• methodology.md · references/
• output-contract.md · references/

SKILL.md (first lines)

# Hypothesis

Satisfy a `product_hypothesis` responsibility by producing `hypothesis_statement`.

## Inputs

Require a target user/segment, an opportunity/problem, and a proposed intervention or belief to test. Prefer an `opportunity_map` and relevant evidence when available.

## Procedure

1. Read [references/methodology.md](references/methodology.md).
2. Express one falsifiable belief linking intervention, target, expected user behavior/outcome, measure, and decision horizon.
3. Separate user behavior from internal business metric where possible.
4. Define a primary measure plus secondary or guardrail measures only when useful and observable.
5. Identify consequential assumptions across value, usability, viability, and feasibility.
6. Define the evidence need and a bounded validation approach, preferring the lowest-cost sufficient evidence source. If empirical work is material, apply `../using-sensemaking/references/experiment-economy-v1.md` before recommending an experiment.
7. Preserve explicit success, pivot, and kill criteria before results exist.
8. Cite supporting evidence and label unsupported premises as assumptions.
9. Render [references/output-contract.md](references/output-contract.md) and return control.

## Stop or downgrade

- If success cannot be falsified under any result, rewrite the hypothesis before returning it.
- If baseline or target values are unknown, keep them unknown or explicitly proposed; do not present invented numbers as observations.
- Do not claim the hypothesis passed or failed without actual experiment or operational evidence.
- Do not launch the test without separate external-action authority when the test changes external state.

## Boundary

```text
hypothesis != experiment warrant
validation approach != experiment requirement
validation approach != validation evidence
hypothesis card != experiment result
```

This Skill may describe what evidence would discriminate the hypothesis, but Experiment Economy owns whether an experiment is warranted.

From ThorStarlord/sensemaking-skills, MIT.

More from ThorStarlord/sensemaking-skills