1 min read228 words

追踪与归因

没有追踪就没有优化——每一分广告费都要有据可查。

追踪体系总览

UTM 参数

"""
UTM 链接生成器
"""
from dataclasses import dataclass
from urllib.parse import urlencode
@dataclass
class UTMBuilder:
"""UTM 参数构建"""
base_url: str
source: str       # 来源: google, facebook, newsletter
medium: str       # 媒介: cpc, social, email
campaign: str     # 活动名
term: str = ""    # 关键词 (搜索广告)
content: str = "" # 区分素材 (A/B 测试)
def build(self) -> str:
params = {
"utm_source": self.source,
"utm_medium": self.medium,
"utm_campaign": self.campaign,
}
if self.term:
params["utm_term"] = self.term
if self.content:
params["utm_content"] = self.content
return f"{self.base_url}?{urlencode(params)}"
# 示例
urls = [
UTMBuilder("https://shop.com/sale", "google", "cpc",
"spring_sale", term="运动鞋", content="ad_v1"),
UTMBuilder("https://shop.com/sale", "facebook", "paid_social",
"spring_sale", content="carousel_red"),
UTMBuilder("https://shop.com/sale", "newsletter", "email",
"spring_sale", content="subject_a"),
]
print("=== UTM 链接生成 ===")
for utm in urls:
print(f"\n来源: {utm.source} / {utm.medium}")
print(f"  {utm.build()}")

归因模型

graph LR subgraph 用户旅程 T1[Day 1: Google 搜索] --> T2[Day 3: Facebook 广告] T2 --> T3[Day 5: 邮件营销] T3 --> T4[Day 7: 直接访问购买] end style T1 fill:#e3f2fd,stroke:#1565c0 style T4 fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px

"""
归因模型对比
"""
ATTRIBUTION_MODELS = {
"末次点击": {
"逻辑": "100% 归因给最后一次点击",
"上例归因": {"Google": "0%", "Facebook": "0%", "邮件": "0%", "直接": "100%"},
"优点": "简单明确",
"缺点": "忽略前期触点",
"适合": "短决策周期产品",
},
"首次点击": {
"逻辑": "100% 归因给第一次接触",
"上例归因": {"Google": "100%", "Facebook": "0%", "邮件": "0%", "直接": "0%"},
"优点": "重视拉新渠道",
"缺点": "忽略转化渠道",
"适合": "品牌曝光考核",
},
"线性归因": {
"逻辑": "平均分配给所有触点",
"上例归因": {"Google": "25%", "Facebook": "25%", "邮件": "25%", "直接": "25%"},
"优点": "公平全面",
"缺点": "无差异化",
"适合": "多触点协同",
},
"时间衰减": {
"逻辑": "越接近转化的触点权重越大",
"上例归因": {"Google": "10%", "Facebook": "20%", "邮件": "30%", "直接": "40%"},
"优点": "符合直觉",
"缺点": "低估种草渠道",
"适合": "促销活动",
},
"数据驱动": {
"逻辑": "AI 根据数据动态分配权重",
"上例归因": {"Google": "35%", "Facebook": "30%", "邮件": "25%", "直接": "10%"},
"优点": "最准确",
"缺点": "需大量数据 (3万+转化)",
"适合": "成熟广告主",
},
}
print("=== 归因模型对比 ===")
for model, info in ATTRIBUTION_MODELS.items():
print(f"\n【{model}】")
print(f"  逻辑: {info['逻辑']}")
print(f"  归因: {info['上例归因']}")
print(f"  适合: {info['适合']}")

Pixel 追踪设置

"""
广告 Pixel 事件映射
"""
PIXEL_EVENTS = {
"Meta Pixel": {
"标准事件": [
{"事件": "PageView", "触发": "所有页面加载", "用途": "构建受众"},
{"事件": "ViewContent", "触发": "商品详情页", "用途": "兴趣追踪"},
{"事件": "AddToCart", "触发": "加入购物车", "用途": "再营销"},
{"事件": "InitiateCheckout", "触发": "进入结算", "用途": "挽回流失"},
{"事件": "Purchase", "触发": "完成付款", "用途": "转化优化"},
],
"回传窗口": "7天点击 / 1天浏览",
},
"Google Ads": {
"标准事件": [
{"事件": "page_view", "触发": "页面加载", "用途": "基础追踪"},
{"事件": "view_item", "触发": "查看商品", "用途": "兴趣追踪"},
{"事件": "add_to_cart", "触发": "加购", "用途": "再营销"},
{"事件": "begin_checkout", "触发": "开始结算", "用途": "漏斗分析"},
{"事件": "purchase", "触发": "购买完成", "用途": "转化出价"},
],
"回传窗口": "30天点击 / 1天浏览",
},
}
print("=== Pixel 事件配置 ===")
for platform, config in PIXEL_EVENTS.items():
print(f"\n【{platform}】 回传窗口: {config['回传窗口']}")
for event in config["标准事件"]:
print(f"  {event['事件']}: {event['触发']} → {event['用途']}")

graph TD OLD[传统: 第三方 Cookie] -->|被淘汰| PROBLEM[追踪失效] PROBLEM --> S1[第一方数据策略] PROBLEM --> S2[服务端追踪 CAPI] PROBLEM --> S3[Google Privacy Sandbox] PROBLEM --> S4[上下文定向] S1 --> CRM[CRM 数据 + 邮件] S2 --> SERVER[服务器端事件回传] S3 --> TOPICS[Topics API] S4 --> CONTEXT[内容语境匹配] style OLD fill:#fce4ec,stroke:#c62828 style S1 fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px style S2 fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px

方案	说明	准确度	实施难度
第一方数据	CRM + 邮件 + 会员体系	最高	中
服务端追踪 CAPI	后端直传广告平台	高	高
Privacy Sandbox	Google Topics API	中	低
上下文定向	根据页面内容投放	中	低
增强归因	Google 机器学习估算	中	低

小结

UTM 参数是追踪的基础，每条广告链接都要加
五种归因模型各有适用场景，大广告主用数据驱动
Pixel 标准事件覆盖完整购买漏斗
Cookie 消亡后，第一方数据 + 服务端追踪是王道

下一章: 预算与 ROI

追踪与归因

追踪体系总览

UTM 参数

归因模型

Pixel 追踪设置

Cookie 消亡与应对

小结