Files
nexus/wiki/concepts/ABTesting.md

106 lines
3.0 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
title: "A/B Testing Framework"
type: concept
tags: ["optimization", "statistics", "ppc", "creative"]
---
## Definition
A/B Testing FrameworkA/B 测试框架)是创意优化的标准方法论,通过对照实验验证假设,区分真实效果提升与随机波动,以数据驱动决策。
## Core Principles
1. **假设驱动Hypothesis-Driven**:每个测试始于明确假设
2. **控制变量Single Variable**:每次只改变一个变量
3. **统计显著性Statistical Significance**:基于置信区间判断结果可靠性
4. **可重复性Reproducibility**:测试结果可推广至更大规模
## Test Design
### Hypothesis Template
> "If [change], then [expected outcome], because [reason]."
示例:
> "If we add urgency language to headlines, then CTR will increase by 10%, because scarcity drives action."
### Sample Size Calculation
| 转化率 | 最小样本(每变体) | 预估测试周期 |
|--------|-------------------|-------------|
| 5% | 2,500 | 7-14 天 |
| 2% | 6,500 | 14-28 天 |
| 1% | 13,000 | 28-56 天 |
**公式**(简化版):
```
n = 16 × σ² / δ²
其中:σ = 标准差,δ = 最小可检测差异
```
### Statistical Significance
| 置信度 | Z-score | 可靠性 |
|--------|---------|--------|
| 90% | 1.645 | 初步参考 |
| 95% | 1.96 | 标准基准 ✅ |
| 99% | 2.576 | 高确定性 |
## Testing Workflow
```
1. Define Hypothesis → 2. Design Test → 3. Launch → 4. Monitor → 5. Analyze → 6. Scale/Iterate
```
### Step 1: Define Hypothesis
- 明确要测试的变量Headline A vs Headline B
- 设定预期提升目标
- 确定主要指标CTR/CVR/CPA
### Step 2: Design Test
- 流量分配50/50 或 80/20
- 测试持续时间2-4 周)
- 样本量计算
### Step 3: Launch
- 确保变体间随机分配
- 记录测试开始时间
- 不在测试期间修改其他变量
### Step 4: Monitor
- 每日检查基本数据
- 避免提前终止(除非严重错误)
- 监控外部因素(季节性/节假日)
### Step 5: Analyze
- 计算统计显著性
- 分析次级指标CVR/CPA
- 撰写结论报告
### Step 6: Scale/Iterate
- 胜出方案规模化推广
- 败出方案归档(积累学习)
- 从败出中提取新假设
## Common Test Types
| 类型 | 测试内容 | 适用场景 |
|------|---------|---------|
| Headline Test | 不同标题变体 | RSA 优化 |
| CTA Test | 不同行动号召 | 转化率优化 |
| Image Test | 不同图片/颜色 | Display/Social |
| Landing Page Test | 不同落地页 | 转化路径优化 |
| Audience Test | 不同受众 | 定向策略优化 |
## Success Criteria
- **统计显著性**95%+ 置信度
- **测试周期**2-4 周
- **最小样本**:每变体至少 1000+ 转化
- **Winner Criteria**显著优于控制组10%+
## Sources
- [[paid-media-creative-strategist]]
- [[paid-media-ppc-strategist]]