intelligems-test-debrief

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/test-debrief

Post-mortem analysis for any A/B test outcome. Extracts learnings, customer behavior insights, and specific next-test suggestions from funnel data and segment patterns.

Works with both active and ended tests. Most useful for tests that have reached a verdict.

针对任意A/B测试结果的事后分析。从漏斗数据和细分群体模式中提炼经验总结、客户行为洞察以及具体的后续测试建议。

支持活跃中和已结束的测试。对已得出结论的测试效果最佳。

Step 0: Workspace Check

步骤0：工作区检查

bash

test -d ~/intelligems-analytics/venv && test -f ~/intelligems-analytics/ig_client.py && echo "READY" || echo "NEEDS_SETUP"

If NEEDS_SETUP: Run the

/intelligems-core

skill first.

bash

test -d ~/intelligems-analytics/venv && test -f ~/intelligems-analytics/ig_client.py && echo "READY" || echo "NEEDS_SETUP"

**如果显示NEEDS_SETUP：**请先运行

/intelligems-core

技能。

Step 1: Get API Key

步骤1：获取API密钥

Check for existing key and ask if missing. Same pattern as other skills.

检查是否已有密钥，若缺失则提示用户。流程与其他技能一致。

Step 2: Copy Debrief Script

步骤2：复制事后分析脚本

bash

cp references/debrief.py ~/intelligems-analytics/debrief.py

bash

cp references/debrief.py ~/intelligems-analytics/debrief.py

Step 3: Select Test

步骤3：选择测试

Pass a test ID directly or let the script list active experiments.

For debriefs on ended tests (the most common use), the user should provide the test ID:

bash

python3 debrief.py <test_id>

直接传入测试ID，或让脚本列出当前活跃的实验。

对于已结束测试的事后分析（最常见场景），用户需提供测试ID：

bash

python3 debrief.py <test_id>

Step 4: Run Analysis

步骤4：运行分析

bash

cd ~/intelligems-analytics && source venv/bin/activate && python3 debrief.py [optional_test_id]

The script will:

Fetch test details + overview analytics
Fetch all 3 segment types (device, visitor, source)
Analyze funnel stages for patterns
Generate customer behavior insights from segment data
Build a structured post-mortem with actionable next steps

bash

cd ~/intelligems-analytics && source venv/bin/activate && python3 debrief.py [optional_test_id]

脚本将执行以下操作：

获取测试详情及概览分析数据
获取全部3种细分类型数据（设备、访客、来源）
分析漏斗各阶段的模式
从细分群体数据中生成客户行为洞察
生成包含可执行后续步骤的结构化事后分析报告

Step 5: Present Debrief

步骤5：展示事后分析结果

Read the output and present conversationally. Structure:

读取输出内容并以对话式方式呈现。结构如下：

1. What Happened

1. 测试结果概述

The verdict and key metrics — one paragraph summary of the test outcome.

测试结论及关键指标——用一段话总结测试结果。

2. Why It Happened — Funnel Analysis

2. 结果成因——漏斗分析

Which funnel stages drove the result? Where did behavior diverge?

哪些漏斗阶段主导了结果？用户行为在何处出现差异？

3. Why It Happened — Segment Patterns

3. 结果成因——细分群体模式

Which segments responded differently? Any contradictions?

哪些细分群体的反应存在差异？是否有矛盾点？

4. Customer Behavior Insights

4. 客户行为洞察

Auto-generated observations. Present these as insights, not raw data:

"Mobile users responded 3x stronger than desktop"
"New visitors drove most of the lift — returning visitors were flat"
"Direct traffic saw no effect, but organic search visitors loved it"

自动生成的观察结论。需将这些内容作为洞察结论展示，而非原始数据：

"移动端用户的反应比桌面端强烈3倍"
"新访客是提升效果的主要驱动群体——老访客数据无明显变化"
"直接流量无显著影响，但自然搜索访客对测试内容接受度高"

5. What to Test Next

5. 后续测试建议

Specific, actionable suggestions based on the debrief findings — not generic advice.

基于事后分析结果提出的具体、可执行建议——而非通用建议。

Step 6: Set Up Slack Automation (Optional)

步骤6：设置Slack自动化（可选）

bash

cd ~/intelligems-analytics && source venv/bin/activate && python3 debrief.py <test_id> --slack "<webhook_url>"

bash

cd ~/intelligems-analytics && source venv/bin/activate && python3 debrief.py <test_id> --slack "<webhook_url>"

Notes

注意事项

Best for ended tests — Most debriefs happen after a test concludes, but it works for active tests too.
5 API calls — 1 detail + 1 overview + 3 segment types.
Insights are auto-generated — The script compares segment performance to find noteworthy patterns without manual inspection.
COGS awareness: Uses Gross Profit per Visitor when COGS data exists.

最适用于已结束的测试——大多数事后分析在测试结束后进行，但该工具也支持活跃测试。
5次API调用——1次获取详情 +1次获取概览 +3次获取细分群体类型数据。
洞察结论自动生成——脚本会对比细分群体的表现，无需人工检查即可发现值得关注的模式。
**COGS感知能力：**当存在COGS数据时，会使用“每位访客的毛利润”指标。