用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/tomevault-io/skills-registry --skill ab-testing命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
| name | ab-testing |
| description | >- Use when this capability is needed. |
Stack: Next.js 16 + Edge Middleware + GA4/GTM + Supabase + Rust-WASM (for stats) Why DIY?: Google Optimize sunset Sept 2023. GA4 has no native A/B testing.
// middleware.ts - Server-side assignment (no flicker)
import { NextResponse, type NextRequest } from 'next/server'
export function middleware(request: NextRequest) {
const response = NextResponse.next()
if (!request.cookies.get('exp_hero')) {
const variant = Math.random() < 0.5 ? 'A' : 'B'
response.cookies.set('exp_hero', variant, { maxAge: 60*60*24*30, path: '/' })
}
return response
}
// Track with GA4
window.dataLayer?.push({
event: 'experiment_view',
experiment_name: 'hero_test',
experiment_variant: variant
})
┌─────────────────────────────────────────────────────────────────────┐
│ A/B TESTING FLOW │
├─────────────────────────────────────────────────────────────────────┤
│ │
│ 1. ASSIGN (Edge Middleware) │
│ ══════════════════════════ │
│ Request → Check cookie → Random assign → Set cookie → Response │
│ ✓ No flicker (server-side) ✓ Consistent (cookie-based) │
│ │
│ 2. RENDER │
│ ═════════ │
│ Server/Client Component → Read cookie → Show variant │
│ │
│ 3. TRACK (GTM + GA4) │
│ ════════════════════ │
│ dataLayer.push → GTM triggers → GA4 events with variant param │
│ │
│ 4. ANALYZE │
│ ═════════ │
│ GA4 Explorations OR Supabase + Rust-WASM Bayesian analysis │
│ │
│ 5. PERSONALIZE (Advanced) │
│ ═════════════════════════ │
│ Contextual bandit → Best variant per user segment │
│ │
└─────────────────────────────────────────────────────────────────────┘
experiment_variant as Custom Dimension in GA4| Method | Best For | Decision Output |
|---|---|---|
| Frequentist | Fixed sample, strict control | p-value < 0.05 → significant |
| Bayesian | Continuous monitoring, intuitive | P(B > A) = 96% → B likely better |
| Multi-Armed Bandit | Optimize during test | Auto-shift traffic to winner |
| Contextual Bandit | Personalization | Best variant per user segment |
Quick Bayesian (Beta-Binomial):
# A: 50/1000 conversions, B: 72/1000
import scipy.stats as stats
a_samples = stats.beta(51, 951).rvs(100000) # Beta(1+50, 1+950)
b_samples = stats.beta(73, 929).rvs(100000) # Beta(1+72, 1+928)
p_b_wins = (b_samples > a_samples).mean() # → ~0.96 (96%)
Full analysis guide: STATISTICAL-ANALYSIS.md
// middleware.ts
const EXPERIMENTS = {
hero_cta: { weight: 0.5 },
pricing_layout: { weight: 0.5 },
signup_flow: { weight: 0.2 }, // 20% on new variant
}
for (const [name, config] of Object.entries(EXPERIMENTS)) {
if (!request.cookies.get(`exp_${name}`)) {
const variant = Math.random() < config.weight ? 'B' : 'A'
response.cookies.set(`exp_${name}`, variant, { maxAge: 2592000, path: '/' })
}
}
// Config stored in Supabase or Vercel Edge Config
const rolloutPhases = {
early_access: 0.1, // 10% new
public_beta: 0.5, // 50% new
general: 1.0 // 100% new (winner)
}
// Include variant in ALL relevant events
window.dataLayer?.push({
event: 'sign_up',
method: 'google',
experiment_name: 'hero_cta',
experiment_variant: getCookie('exp_hero_cta'),
eventId: crypto.randomUUID() // De-duplication
})
| Use Case | Why Rust |
|---|---|
| Monte Carlo simulation (100k+ draws) | 10-100x faster than JS |
| Bayesian posterior computation | Numerical precision |
| Contextual bandit inference | Real-time ML at edge |
| Cross-platform consistency | Same logic in browser + server |
WASM is NOT needed for: Simple random assignment, cookie handling, event tracking
See: RUST-WASM.md
| Don't | Why |
|---|---|
| Client-side variant assignment | Causes flicker, inconsistent |
| End test early ("B winning after 2 days!") | Random noise, not signal |
| Multiple changes in one variant | Can't isolate what worked |
| Overlapping tests on same element | Interaction effects confound |
| Skip sample size calculation | Under-powered = false negatives |
| Ignore segments | Winner overall may lose for key segment |
experiment_variant registered in GA4| I need to... | Read |
|---|---|
| Implement variant assignment | VARIANT-ASSIGNMENT.md |
| Choose a statistical method | STATISTICAL-ANALYSIS.md |
| Set up GTM/GA4 tracking | GA4-GTM-TRACKING.md |
| Build admin dashboard | ADMIN-DASHBOARD.md |
| Add personalization/bandits | PERSONALIZATION.md |
| Optimize with Rust-WASM | RUST-WASM.md |
| Quick lookup (tools, formulas) | QUICK-REFERENCE.md |
| Topic | Reference |
|---|---|
| Server-side assignment, multiple experiments, weighted splits | VARIANT-ASSIGNMENT.md |
| Frequentist vs Bayesian vs Bandits, sample size, pitfalls | STATISTICAL-ANALYSIS.md |
| GTM variables, GA4 events, BigQuery queries, debugging | GA4-GTM-TRACKING.md |
| Database schema, API routes, UI components, real-time updates | ADMIN-DASHBOARD.md |
| Segments, rules-based, Thompson sampling, contextual bandits | PERSONALIZATION.md |
| Rust setup, beta sampling, bandit implementation, Next.js integration | RUST-WASM.md |
| Tool recommendations, decision guide, formulas, cheat sheet | QUICK-REFERENCE.md |
Source: danzam98/claude-skills-toolkit — distributed by TomeVault.
基于 SOC 职业分类