Experiment Designer
Experiment Designer helps plan, prioritize, and evaluate product experiments with statistical rigor. It guides hypothesis writing, sample size estimation, ICE prioritization, and result interpretation, producing clear experiment plans and defensible decisions.
✨ What it does
- Writes hypotheses in If/Then/Because format.
- Defines primary, guardrail, and secondary metrics.
- Estimates sample size via a Python calculator script.
- Prioritizes experiments with ICE scoring.
- Provides a hypothesis quality checklist and common pitfalls.
- Offers statistical interpretation guardrails.
🎯 When to use it
- When planning A/B or multivariate tests and need to define hypotheses and success metrics.
- When estimating required sample size or minimum detectable effect for an experiment.
- When prioritizing a backlog of experiment ideas using ICE scoring.
- When interpreting statistical results to make product decisions.
🚀 How to use
Trigger by asking for help designing an experiment, writing a hypothesis, or planning sample size. Provide context like the intervention, target metric, baseline rate, and expected effect. The skill will guide you through the workflow and may use the sample size calculator script.
Example prompts:
Help me design an A/B test for a new onboarding flow. Baseline conversion is 12%, and we want to detect a 2% absolute increase.
Write a hypothesis for changing the pricing page layout and estimate sample size.
📄 Output: A structured experiment plan including hypothesis, metrics, sample size, prioritization score, and interpretation guidance.
📦 Add this skill to Claude Code
# 1. Get the skills repo
git clone --depth 1 https://github.com/alirezarezvani/claude-skills /tmp/claude-skills
# 2. Copy this skill into your project (or ~/.claude/skills for all projects)
mkdir -p .claude/skills
cp -r /tmp/claude-skills/product-team/skills/experiment-designer .claude/skills/experiment-designerSkill source: product-team/skills/experiment-designer/SKILL.md
⚠️ Good to know
Requires Python to run the sample size calculator script; assumes user has baseline metrics and statistical knowledge.
❓ FAQ
What is the recommended hypothesis format?
Use If/Then/Because: If we change [intervention], Then [metric] will change by [expected direction/magnitude], Because [behavioral mechanism].
How do I prioritize experiments?
Use ICE scoring: Impact * Confidence * Ease divided by 10, where each factor is rated on a scale.
What should I do if my test is underpowered?
Increase sample size, relax the minimum detectable effect, or accept higher risk of false negatives; the skill helps estimate required sample size.
🤖 Overview, features, install steps and FAQ were generated from the project's SKILL.md on Sep 4, 2026. Always check the original source before running commands.