GA4 Session_starts未触发问题求助及BigQuery解决方案咨询

一、可能的触发失败原因 #

• 跨域名跟踪配置遗漏：多域名场景下未启用GA4跨域名测量，用户跨域名跳转时会话未被正确延续，导致session_start不触发，但后续购买等事件仍可被记录。
• 代码加载异常：部分页面的GA4初始化代码被广告拦截器拦截、加载延迟或存在语法错误，页面加载阶段未触发session_start，但后续交互事件（如购买）仍能触发。
• 会话参数被篡改：自定义代码或第三方工具修改了GA4的session_id等核心会话参数，系统误判为已有会话，跳过session_start触发。
• 非交互事件抢占触发顺序：用户进入页面后，非交互事件（如自定义埋点）先于page_view加载完成，导致GA4未识别为新会话。

二、BigQuery解决方案 #

1. 定位缺失Session Starts的会话 #

先筛选出有购买行为但无session_start的会话，确认受影响范围：

WITH session_summary AS (
  SELECT
    user_pseudo_id,
    session_id,
    ARRAY_AGG(event_name) AS event_list,
    MAX(IF(event_name = 'purchase', 1, 0)) AS has_purchase,
    MIN(event_timestamp) AS first_event_time
  FROM `your-project.analytics_xxxxxx.events_*`
  WHERE _TABLE_SUFFIX BETWEEN FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)) 
                          AND FORMAT_DATE('%Y%m%d', CURRENT_DATE())
  GROUP BY user_pseudo_id, session_id
)
SELECT * FROM session_summary
WHERE has_purchase = 1 AND NOT 'session_start' IN UNNEST(event_list)

2. 虚拟补全Session Starts事件 #

基于会话的首个事件时间戳，生成对应的session_start记录，合并到原始数据中：

WITH raw_events AS (
  SELECT * FROM `your-project.analytics_xxxxxx.events_*`
  WHERE _TABLE_SUFFIX BETWEEN FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)) 
                          AND FORMAT_DATE('%Y%m%d', CURRENT_DATE())
),
missing_sessions AS (
  SELECT
    user_pseudo_id,
    session_id,
    MIN(event_timestamp) AS session_start_ts,
    MAX(traffic_source.source) AS source,
    MAX(traffic_source.medium) AS medium,
    MAX(page_location) AS first_page
  FROM raw_events
  WHERE session_id IN (
    SELECT session_id
    FROM (
      SELECT
        user_pseudo_id,
        session_id,
        ARRAY_AGG(event_name) AS event_list,
        MAX(IF(event_name = 'purchase', 1, 0)) AS has_purchase
      FROM raw_events
      GROUP BY user_pseudo_id, session_id
    )
    WHERE has_purchase = 1 AND NOT 'session_start' IN UNNEST(event_list)
  )
  GROUP BY user_pseudo_id, session_id
),
virtual_session_starts AS (
  SELECT
    user_pseudo_id,
    session_id,
    'session_start' AS event_name,
    session_start_ts AS event_timestamp,
    source,
    medium,
    first_page,
    CURRENT_TIMESTAMP() AS created_at
  FROM missing_sessions
)
-- 合并原始数据与虚拟会话起始事件
SELECT * FROM raw_events
UNION ALL
SELECT
  user_pseudo_id,
  session_id,
  event_name,
  event_timestamp,
  NULL AS event_value_in_usd,
  NULL AS event_previous_timestamp,
  NULL AS event_params,
  NULL AS items,
  source,
  medium,
  NULL AS campaign,
  NULL AS term,
  NULL AS content,
  first_page,
  NULL AS page_referrer,
  NULL AS device_category,
  NULL AS operating_system,
  NULL AS country,
  NULL AS region,
  NULL AS city,
  created_at
FROM virtual_session_starts

3. 修正核心指标计算 #

基于补全后的数据集，重新计算页面浏览量、参与率及来源/媒介数据：

WITH combined_data AS (
  -- 替换为上述合并后的数据集或临时表
  SELECT * FROM `your-project.analytics_xxxxxx.combined_events_30d`
)
-- 来源/媒介会话与购买统计
SELECT
  traffic_source.source,
  traffic_source.medium,
  COUNT(DISTINCT CONCAT(user_pseudo_id, '-', session_id)) AS total_sessions,
  SUM(IF(event_name = 'purchase', 1, 0)) AS total_purchases
FROM combined_data
GROUP BY source, medium
ORDER BY total_sessions DESC;

-- 页面浏览量与参与率
SELECT
  page_location,
  COUNT(DISTINCT CONCAT(user_pseudo_id, '-', session_id)) AS sessions,
  COUNT(IF(event_name = 'page_view', 1, 0)) AS page_views,
  ROUND(
    COUNT(DISTINCT CASE WHEN event_name IN ('page_view', 'purchase', 'scroll') THEN CONCAT(user_pseudo_id, '-', session_id) END) 
    / COUNT(DISTINCT CONCAT(user_pseudo_id, '-', session_id)),
    4
  ) AS engagement_rate
FROM combined_data
GROUP BY page_location
ORDER BY sessions DESC;

内容的提问来源于stack exchange，提问作者Max_Stone