搜索技能.../

Co Growth Experiment Design | Skills Pool

技能档案

Co Growth Experiment Design

Experiment design skill. Hypothesis-driven A/B test planning with sample size calculation, MDE definition, guardrail metrics, and runtime estimation. Use when DESIGNING A/B tests, calculating sample sizes, defining hypotheses, setting guardrail metrics, or planning experiment rollout.

nahisaho0 星标2026年3月30日

职业
分类: 测试

技能内容

Experiment Design

Hypothesis-driven A/B test design with statistical rigor.

Use This Skill When

Designing a new A/B test or multivariate experiment.
Formulating a testable hypothesis with success criteria.
Calculating required sample size given MDE and baseline.
Defining guardrail metrics that must not degrade.
Estimating experiment runtime.

Workflow

Formulate hypothesis: "If [change], then [metric] will [direction] by [MDE]."
Define metrics: primary, secondary, and guardrail.
Calculate sample size: inputs = baseline rate, MDE, α=0.05, β=0.20.
Estimate runtime: sample size ÷ daily traffic × seasonality buffer.
Document randomization unit, allocation ratio, and exclusion criteria.
Save experiment plan using template.

Deliverables

: experiment rationale and design summary.

相关技能

快速安装

Co Growth Experiment Design

npx skills add nahisaho/coreclaw-marketplace

下载 Skill 打开源码仓库

作者: nahisaho
星标: 0
更新时间: 2026年3月30日
职业

本页内容

01Experiment Design

report.md

results/experiment-plan.md: full plan using template.

results/sample-size-calc.md: calculation with all inputs.

Quality Gates

Hypothesis is falsifiable with a clear primary metric.
MDE is business-justified (not just "detectable").
Sample size uses α=0.05, β=0.20 (or justified alternatives).
At least one guardrail metric is defined.
Runtime accounts for weekly seasonality (≥2 full weeks).

If any gate fails: identify the issue, fix, and re-validate.

Gotchas

An MDE of "any improvement" is not valid. Require a minimum business-meaningful threshold (e.g., +2pp conversion).
Experiments shorter than 2 weeks miss weekly seasonality cycles and produce biased estimates.
Guardrail metrics added after launch are post-hoc and invalidate the pre-registered analysis plan.

Validation Loop

Generate experiment design
Check: falsifiable hypothesis, sample size calc, guardrails, runtime ≥ 2 weeks
If any check fails → revise design
User approval before launch ⏸️

02

Use This Skill When

05Quality Gates

07Validation Loop

市场研究分析师与营销专家

Test

Use when you need to run tests for React core. Supports source, www, stable, and experimental channels.

Feature Flags

Use when feature flag tests fail, flags need updating, understanding @gate pragmas, debugging channel-specific test failures, or adding new flags to React.

Unit Tests

Use when running unit tests in the VS Code repo. Covers the runTests tool, scripts/test.sh (macOS/Linux) and scripts/test.bat (Windows), and their supported arguments for filtering, globbing, and debugging tests.

microsoft184.0k

Integration Tests

Use when running integration tests in the VS Code repo. Covers scripts/test-integration.sh (macOS/Linux) and scripts/test-integration.bat (Windows), their supported arguments for filtering, and the difference between node.js integration tests and extension host tests.

microsoft184.0k

Write Frontend Tests

Analyze the current branch diff against dev, plan integration tests for changed frontend pages/components, and write them. TRIGGER when user asks to write frontend tests, add test coverage, or 'write tests for my changes'.

Significant-Gravitas183.5k

Golang Testing

テスト駆動開発とGoコードの高品質を保証するための包括的なテスト戦略。

市场研究分析师与营销专家