KPI跟踪器

概述

KPI跟踪器技能提供了定义、计算和监控关键绩效指标的全面能力。它支持从定义到跟踪、预警和报告的完整KPI生命周期，用于商业智能和绩效管理。

能力

KPI公式定义和验证
目标和阈值管理
交通灯状态计算
趋势分析和预测
下钻层次配置
基准比较
方差分析
自动预警生成

使用流程

KPI框架开发
高管仪表板开发
运营报告系统设计

使用方法

KPI定义

# 定义KPI结构
kpi_definition = {
    "name": "客户获取成本",
    "code": "CAC",
    "category": "市场营销",
    "description": "获取新客户的总成本",
    "formula": "total_marketing_spend / new_customers_acquired",
    "unit": "货币",
    "polarity": "越低越好",
    "frequency": "月度",
    "owner": "市场总监",
    "data_sources": [
        {"name": "营销支出", "source": "财务系统", "table": "费用"},
        {"name": "新客户", "source": "客户关系管理系统", "table": "客户"}
    ]
}

目标配置

# 定义目标和阈值
targets = {
    "kpi": "CAC",
    "period": "2024-Q1",
    "target": 150,
    "thresholds": {
        "green": {"max": 150},
        "yellow": {"min": 150, "max": 200},
        "red": {"min": 200}
    },
    "benchmark": {
        "industry_average": 180,
        "best_in_class": 100,
        "previous_period": 175
    }
}

层次配置

# 定义下钻层次
hierarchy = {
    "kpi": "收入",
    "levels": [
        {"name": "总计", "aggregation": "sum"},
        {"name": "区域", "dimension": "地理", "aggregation": "sum"},
        {"name": "产品线", "dimension": "产品", "aggregation": "sum"},
        {"name": "销售代表", "dimension": "销售人员", "aggregation": "sum"}
    ]
}

预警配置

# 配置自动预警
alert_config = {
    "kpi": "CAC",
    "conditions": [
        {
            "type": "阈值突破",
            "threshold": "red",
            "consecutive_periods": 2,
            "notification": ["email", "slack"]
        },
        {
            "type": "趋势",
            "direction": "上升",
            "periods": 3,
            "min_change_percent": 10,
            "notification": ["email"]
        },
        {
            "type": "预测突破",
            "horizon": 3,
            "probability": 0.8,
            "notification": ["email", "dashboard"]
        }
    ]
}

KPI类别

类别	示例KPI
财务	收入、利润率、投资回报率、客户获取成本、客户终身价值
客户	净推荐值、流失率、客户满意度、留存率
运营	周期时间、缺陷率、利用率
增长	月度经常性收入增长、用户增长、市场份额
效率	单位成本、人均收入

输入模式

{
  "operation": "define|calculate|track|alert",
  "kpi_definition": {
    "name": "string",
    "formula": "string",
    "unit": "string",
    "polarity": "higher_is_better|lower_is_better",
    "frequency": "string"
  },
  "targets": {
    "value": "number",
    "thresholds": "object"
  },
  "data": {
    "source": "string",
    "period": "string",
    "values": "object"
  },
  "analysis_options": {
    "trend_analysis": "boolean",
    "forecast": "boolean",
    "variance_analysis": "boolean"
  }
}

输出模式

{
  "kpi_values": {
    "current_value": "number",
    "previous_value": "number",
    "target": "number",
    "variance": "number",
    "variance_percent": "number",
    "status": "green|yellow|red"
  },
  "trend_analysis": {
    "direction": "improving|stable|declining",
    "change_percent": "number",
    "periods_analyzed": "number"
  },
  "forecast": {
    "next_period": "number",
    "confidence_interval": ["number", "number"],
    "will_breach_target": "boolean"
  },
  "drill_down": {
    "dimension_values": "object"
  },
  "alerts": [
    {
      "type": "string",
      "severity": "string",
      "message": "string"
    }
  ]
}

最佳实践

每个仪表板限制5-7个KPI（避免指标过载）
为每个KPI定义明确的所有者
设定SMART目标（具体的、可衡量的、可实现的、相关的、有时限的）
包含领先指标，而不仅仅是滞后指标
与业务利益相关者验证公式
记录数据血统和计算逻辑
定期审查和淘汰过时的KPI

数据质量

该技能验证：

数据完整性（缺失值）
数据新鲜度（最后更新时间）
公式有效性（除以零、空值处理）
合理范围（异常值检测）

集成点

输入决策可视化用于仪表板
连接数据叙事用于叙述
支持时间序列预测器用于预测
集成预警系统用于通知