Agent Skills: Hypothesis Driven 80/20

Hypothesis Driven 80/20

Use $ARGUMENTS as initial context.

When to use this skill

• Ambiguous problems that need fast direction before full analysis.
• Prioritizing limited team capacity across competing hypotheses.
• Designing a test plan with clear stop or continue criteria.
• Converting broad strategic questions into measurable experiments.

When not to use this skill

• Do not use it to create a broad issue tree, a market inventory, or a portfolio allocation without a testable decision question.
• Do not label opinions as hypotheses when no measurable disconfirming signal is possible.

Required inputs

• Governing business question and target metric.
• Time horizon and decision deadline.
• Data constraints and available analysis bandwidth.
• Current baseline, unit of analysis, and minimum reliable signal.

Workflow

1. Define one decision question and explicit success metric.
2. Generate 3-7 falsifiable hypotheses using strict format.
3. For each hypothesis define expected signal, disconfirming signal, kill threshold, and measurement window.
4. Score hypotheses with 80/20 lens: impact, confidence, effort, speed.
5. Select top 1-2 hypotheses for minimum viable tests.
6. Build workplan with owners, deadlines, and decision checkpoints.

Ask-first questions

Ask up to 3 questions before ranking hypotheses:

1. Which single metric determines decision success?
2. What is the latest acceptable date for a go or no-go decision?
3. Which datasets are trusted and immediately accessible?

Assumption policy

• If critical data is unavailable, proceed with transparent assumptions.
• Annotate each assumption with confidence and validation action.
• Avoid merged hypotheses; keep one causal chain per hypothesis.
• Record data quality, sample limitations, and the test's decision relevance.

Output contract

Always produce these sections in order:

1. Context
2. Decision or Recommendation
3. Analysis
4. Risks
5. Next Actions
6. Assumptions

• Every test includes an owner, due date, success signal, and decision checkpoint.
• Label evidence as Fact, Inference, Assumption, or Unknown.

Guardrails

• Hypotheses must follow: "If X, then Y, because Z.".
• Avoid descriptive statements that cannot be disproven.
• Include both confirming and disconfirming signals.
• Stop low-value analysis once kill threshold is reached.
• Do not use false precision when the data cannot support a numeric threshold.

Handoffs

• Use consulting-issue-tree-mece when the causal structure is still unclear.
• Use decision-analysis-under-uncertainty when several tested options must be compared.
• Use execution-operating-system when experiments become a sustained delivery program.

Resources

• references/hypothesis-design.md - Falsifiability rules and signal design.
• references/80-20-prioritization.md - Scoring and sequencing framework.
• templates/hypothesis-plan.md - Decision-ready hypothesis template.
• examples/hypothesis-example.md - Golden example with incomplete inputs.

Keywords

hypothesis driven, 80/20, falsifiable hypothesis, prioritization, test plan, consulting