Applicable Scenarios
- Business data weekly / monthly report
- Special analysis (churn / conversion / growth)
- A/B test conclusion report
- Decision support report
- Data-driven PRD
Core Principles
1. 报告给决策者看
首页有结论和建议
2. 洞察 > 数字
不只列数字,要解释"为什么"和"怎么办"
-
So What? Every finding must answer "so what?"
-
Recommendations must be actionable Not "strengthen operations," but "send a 7-day coupon to churned users"
-
Data must have sources Reproducible (attach SQL / link)
## 报告结构
```text
1. 摘要(30 秒能看完)
2. 背景(为什么做这个分析)
3. 关键发现(3~5 条)
4. 详细分析(图表 + 解读)
5. 建议与行动项
6. 局限性(诚实)
7. 附录(SQL / 数据)
Workflow
1. 明确分析目标(与 PM 对齐)
2. 数据收集(SQL / 日志)
3. 分析(漏斗 / 留存 / 对比)
4. 可视化(图表)
5. 提炼洞察(So What?)
6. 写建议(可行动)
7. 评审(与 PM / 业务方)
8. 沉淀(field-journal)
Supporting Template
templates/analysis-report-template.md— Complete analysis report template
Quality Self-Check
□ 摘要有结论
□ 发现有数据支撑
□ 洞察回答 So What
□ 建议可行动
□ 图表配合文字
□ 局限性诚实
□ SQL 可复现
□ 与决策者评审过
Common Pitfalls
- Listing numbers without explanation——decision-makers cannot understand
- Recommendations too broad——"improve retention" is not actionable
- Data not reproducible——no SQL link
- Not stating limitations——overconfidence
- Report too long——no one reads it
Collaboration with other skills
上游:
sql-analysis → 数据查询
ab-testing → 实验结论
data-visualization → 配图
metric-design → 指标定义
下游:
product-manager → 数据驱动决策
project-manager → 行动项跟踪