experiment-designer · diff
v1.0.0 to git:20260420.513e1d3
55 added, 41 removed. Audit A to A.
---
name: experiment-designer
- description: Designs A/B tests from hypotheses and interprets experiment results
- with statistical rigour. Use when user says "run an experiment", "design an A/B
- test", "test this feature", "interpret these results", "was this experiment
- successful", or "what sample size do I need".
- metadata:
- author: Mohit Aggarwal
- version: 1.0.0
- category: data-and-metrics
- tags: [experimentation, data, analytics, ab-testing]
- documentation: https://github.com/mohitagw15856/pm-claude-skills
+ description: "Design statistically rigorous A/B tests and interpret experiment results. Use when asked to design an experiment, run an A/B test, calculate sample size, interpret test results, or assess whether an experiment was successful. Produces a complete experiment design with hypothesis, sample size, run time, success criteria, and risk flags — or a results interpretation with ship/iterate/kill recommendation."
---
+
# Experiment Designer Skill
- ## Purpose
- Produce rigorous experiment designs from product hypotheses, and interpret
- results with statistical and practical significance — so you can defend every
- decision to a sceptical engineering lead or data scientist.
+ Produce rigorous experiment designs from product hypotheses, and interpret results with statistical and practical significance — so you can defend every decision to a sceptical engineering lead or data scientist.
+ ## Required Inputs
+
+ Ask the user for these if not provided:
+ **For experiment design:**
+ - Hypothesis (what change, what metric, what expected movement)
+ - Current baseline metric value
+ - Minimum detectable effect (MDE) — the smallest lift worth caring about
+ - Available daily sample size
+
+ **For results interpretation:**
+ - Control and variant results (raw numbers or percentages)
+ - P-value or confidence interval
+ - Run duration (days)
+ - Any anomalies observed during the test
+
## Two-Phase Process
### Phase 1: Experiment Design
- **Required inputs:** hypothesis, primary metric, current baseline, minimum
- detectable effect (MDE), available sample size per day.
-
- **Output:**
- - Hypothesis restated as: "If we [change], we expect [metric] to [move by X%]
- because [reason]"
- - Control and variant definitions
- - Primary metric (one only)
- - Secondary guardrail metrics (2-3 max)
- - Required sample size (calculated from MDE and baseline)
- - Estimated run time in days
- - Pre-defined success criteria (before the test runs — no moving goalposts)
- - Design risk flags: novelty effects, seasonal confounds, multiple testing issues,
- network effects, sample ratio mismatch risks
+ 1. Restate hypothesis as: "If we [change], we expect [metric] to [move by X%] because [reason]"
+ 2. Define control and variant clearly
+ 3. Select primary metric (one only) and secondary guardrail metrics (2-3 max)
+ 4. Calculate required sample size from MDE and baseline
+ 5. Estimate run time in days
+ 6. Set pre-defined success criteria before the test runs — no moving goalposts
+ 7. Flag design risks: novelty effects, seasonal confounds, multiple testing issues, network effects, sample ratio mismatch
### Phase 2: Results Interpretation
- **Required inputs:** control results, variant results, p-value or raw numbers,
- run duration, any anomalies observed.
+ 1. Assess statistical significance (p < 0.05 threshold)
+ 2. Assess practical significance: was the lift meaningful for the business, not just real?
+ 3. Interpret confidence intervals
+ 4. Investigate confounding factors
+ 5. Recommend: Ship / Iterate / Kill / Run follow-up test
+ 6. **Validate** — Confirm the test ran for the full planned duration. Flag if it was stopped early (peeking problem). Confirm sample ratio mismatch did not occur.
- **Output:**
- - Statistical significance assessment (p < 0.05 threshold)
- - Practical significance: was the lift meaningful for the business, not just real?
- - Confidence interval interpretation
- - Confounding factors to investigate
- - Recommendation: Ship / Iterate / Kill / Run follow-up test
- - If "Iterate": specific hypotheses to test next
+ ## Output Structure
+ **[Design or Results header based on phase]**
+
+ *Hypothesis:* "If we [change], we expect [metric] to [move by X%] because [reason]"
+
+ *Primary metric:* [One metric only]
+ *Guardrail metrics:* [2-3 max]
+ *Required sample size:* [n per variant]
+ *Estimated run time:* [days]
+ *Pre-defined success threshold:* [specific number]
+ *Design risk flags:* [any concerns]
+
+ **Results (Phase 2 only):**
+ *Statistical significance:* [p-value and conclusion]
+ *Practical significance:* [lift size vs. business threshold]
+ *Recommendation:* Ship / Iterate / Kill / Follow-up — [rationale]
+
## Quality Checks
- - Never interpret results from an underpowered test without flagging it
- - Always distinguish statistical from practical significance
- - Flag if test was stopped early (peeking problem)
- - Note if sample ratio mismatch occurred
+
+ - [ ] Hypothesis specifies the change, the metric, the direction, and the reason
+ - [ ] Primary metric is singular — guardrail metrics are secondary
+ - [ ] Success criteria are defined before the test launches (not after seeing results)
+ - [ ] Test was not stopped early (or flagged clearly if it was)
+ - [ ] Practical significance assessed separately from statistical significance
+ - [ ] Sample ratio mismatch is checked in results interpretation