指标异动诊断
SkillMonitoring & opsDiagnose the causes of metric changes. Reproduce the change, define comparison baselines, validate drivers, distinguish structural factors from mixed effects, and produce executive-ready KPI interpretation. Trigger scenarios: why a metric rose/fell, weekly/monthly report interpretation, diagnose thi
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the 指标异动诊断 skill
What this skill tells your AI
The instructions your AI receives, as published by rongxinzy/rongxinai in SKILLs/zhiyuan-expert-manager/presets/data-analyst/skills/metric-diagnosis/SKILL.md and read by ahel’s review.
目标
回答"指标为什么变了",并给出有证据支撑的驱动因素与决策建议。区分已验证事实、合理假设与业务背景。
工作流
Step 1:复现变化
- 明确指标口径(分子/分母、过滤条件、时间粒度)
- 用数据复现变化幅度与发生时间点
- 记录对比基准(环比 / 同比 / 目标值)
Step 2:定义对比
- 单指标变化:量级、方向、持续性(单点还是趋势)
- 确认是否伴随口径变更、数据回填或异常事件(先用 data-quality-review 排除数据问题)
Step 3:分解驱动
- 结构分解:总量 = 分群 × 渗透 × 频次 × 单价(按业务链条拆)
- 混合效应:区分"结构变化(mix shift)"与"群内变化(within-segment)"
- 对每个候选驱动因素给出贡献度估算与证据
Step 4:验证与结论
- 对主驱动因素做反向验证(剔除后变化是否消失)
- 无法验证的解释标注为假设,不冒充结论
- 输出:什么变了 → 为什么变(已验证/假设分开)→ 业务影响 → 建议动作 → 置信度
输出模板
**What changed**: [指标、幅度、时间]
**Why it changed**: [已验证驱动 + 证据;假设 + 待验证项]
**Business impact**: [对业务意味着什么]
**Confidence**: [高/中/低 + 理由]
**Next step**: [建议动作或跟进项]
工具建议
- 数值计算与分解:
xlsx或database-inspector取数后计算 - 趋势与分解可视化:
code-to-chart(趋势线、堆叠条形图呈现 mix shift) - 周报/月报产出:结合
analytics-report的结构规范
Signals
- GitHub stars
- 151
- Forks
- 3
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
metric-diagnosis- Source
- github.com/rongxinzy/rongxinai