带参数的 Firebase 封闭漏斗答案

【问题标题】：Firebase Closed Funnels With Parameters带参数的 Firebase 封闭漏斗
【发布时间】：2018-11-30 15:16:48
【问题描述】：

我在网上看到过各种Firebase封闭漏斗examples，但都不是参数级别的。

问题：

我想根据 Firebase 事件参数值创建一个封闭的用户漏斗。
更具体地说，我想根据以下 3 个漏斗步骤创建一个封闭的用户漏斗：
- first_open（用户第一次打开应用）
- Tutorial_LessonStarted 带有参数键 LessonNumber 和参数值 1
- Tutorial_LessonStarted 带有参数键 LessonNumber 和参数值 2

（更新）我将原始测试查询移至答案部分。

【问题讨论】：

好吧，至少部分问题是你有e0.event_timestamp > e1_ts，如果那个e0事件是first_open，这可能永远不会是真的。您可能打算使用<
我更新了我的代码。请查看下面的答案部分。

标签： firebase google-bigquery firebase-analytics

【解决方案1】：

SELECT
    e0.event_date,
    COUNT(DISTINCT e0.user_pseudo_id) AS step1_firstOpeners,
    COUNT(DISTINCT e1_user) AS step2_lesson1Started,
    COUNT(DISTINCT e2_user) AS step3_lesson2Started
FROM `projectID.analytics_XXXX.events_*` AS e0
LEFT JOIN (SELECT
                event_name AS e1_eventname,
                e.user_pseudo_id AS e1_user,
                event_timestamp AS e1_ts
           FROM `projectID.analytics_XXXX.events_*` AS e, UNNEST(e.event_params) as eventparams
           WHERE platform = "ANDROID"
                AND event_name = "Tutorial_LessonStarted"
                AND eventparams.key = "LessonNumber"
                AND eventparams.value.string_value  = "1")
    ON e0.user_pseudo_id = e1_user AND e0.event_timestamp < e1_ts AND e1_eventname = "Tutorial_LessonStarted"
LEFT JOIN (SELECT 
                event_name AS e2_eventname,
                e.user_pseudo_id AS e2_user, 
                event_timestamp AS e2_ts
           FROM `projectID.analytics_XXXX.events_*` AS e, UNNEST(e.event_params) as eventparams
           WHERE platform = "ANDROID"
                AND event_name = "Tutorial_LessonStarted" 
                AND eventparams.key = "LessonNumber"
                AND eventparams.value.string_value  = "2")
    ON e1_user = e2_user AND e2_ts > e1_ts AND e2_eventname = "Tutorial_LessonStarted" 

WHERE 
    e0.event_name = 'first_open' 
    AND user_first_touch_timestamp >= 1541808000000000 AND user_first_touch_timestamp < 1541894400000000 -- 10 November 2018
    AND _TABLE_SUFFIX = '20181110'
AND platform = "ANDROID"
GROUP BY 1
ORDER BY 1 DESC

【讨论】：