Designs and implements statistically valid A/B tests to optimize product performance and conversion rates.
This skill empowers Claude to act as an experimentation expert, guiding users through the end-to-end process of setting up scientific split tests. It provides structured frameworks for creating data-driven hypotheses, calculating required sample sizes to avoid the 'peeking problem,' and selecting meaningful primary, secondary, and guardrail metrics. Whether you are testing UI changes, copy variations, or backend logic, this skill ensures your experiments produce actionable, statistically significant results while avoiding common pitfalls like underpowered tests or misinterpreted data.
主な機能
01Implementation strategies for client-side and server-side testing
02Post-test analysis checklists for statistical significance and confidence
03Structured hypothesis framework based on data and observations
040 GitHub stars
05Statistical power and sample size calculation guidance
06Selection of primary, secondary, and guardrail metrics
ユースケース
01Setting up guardrail metrics to ensure a pricing change doesn't increase churn
02Designing a headline split test to increase landing page conversion
03Calculating the required traffic and duration for a new feature experiment
