Merge pull request #14 from TechArtists/multiple-project-source

add session_span

Author: SorinSV

models/analytics/google_analytics_installs.sql

@@ -23,15 +23,31 @@
 {%- set columnsForEventDimensions = tmp_res[0] -%}
 {%- set eventDimensionsUnnestedCount = tmp_res[1]  -%}
 
+{%- set install_types = [
+    "custom_event",
+    "our_first_open",
+    "first_open",
+    "any_first_event"
+] -%}
+
 WITH data as (
     SELECT    DATE(event_ts) as event_date
             , DATE(install_ts) as install_date
             , project_id
             , dataset_id
             , install_age
             , {{ ta_firebase.unpack_columns_into_minicolumns(columnsForEventDimensions, miniColumnsToIgnoreInGroupBy, [], "", "") }}
-            , COUNT(1) as cnt
-            , COUNT(DISTINCT(user_pseudo_id)) as users
+            , COUNT(1) AS cnt
+            , COUNT(IF(has_post_install_events_after_10s, 1, NULL)) AS cnt_qualified
+            , COUNT(DISTINCT user_pseudo_id) AS users
+            , COUNT(DISTINCT IF(has_post_install_events_after_10s, user_pseudo_id, NULL)) AS users_qualified
+
+        {% for install_type in install_types %}
+            , COUNT(IF(user_pseudo_id_{{ install_type }}, 1, NULL)) AS cnt_{{ install_type }}
+            , COUNT(IF(user_pseudo_id_{{ install_type }} AND has_post_install_events_after_10s, 1, NULL)) AS cnt_{{ install_type }}_qualified
+            , COUNT(DISTINCT IF(user_pseudo_id_{{ install_type }}, user_pseudo_id, NULL)) AS users_{{ install_type }}
+            , COUNT(DISTINCT IF(user_pseudo_id_{{ install_type }} AND has_post_install_events_after_10s, user_pseudo_id, NULL)) AS users_{{ install_type }}_qualified
+        {% endfor %}
 
     FROM {{ ref("google_analytics_installs_raw") }}
     WHERE {{ ta_firebase.analyticsDateFilterFor('event_date') }}
@@ -44,5 +60,13 @@ SELECT  event_date
       , install_age
       , {{ ta_firebase.pack_minicolumns_into_structs_for_select(columnsForEventDimensions, miniColumnsToIgnoreInGroupBy, "", "") }}
       , cnt
+      , cnt_qualified
       , users
-FROM data
+      , users_qualified
+{% for install_type in install_types %}
+      , cnt_{{ install_type }}
+      , cnt_{{ install_type }}_qualified
+      , users_{{ install_type }}
+      , users_{{ install_type }}_qualified
+{% endfor %}
+FROM data

models/analytics/google_analytics_installs_raw.sql

@@ -7,7 +7,7 @@
      },
     incremental_strategy = 'insert_overwrite',
     require_partition_filter = true,
-    cluster_by = ["platform", "bundle_id"],
+    cluster_by = ["platform", "bundle_id"]
 ) }}
     -- incremental_strategy='insert_overwrite',
     -- require_partition_filter = false
@@ -26,7 +26,7 @@ WITH  custom_install_event AS (
 )
 , ta_install_event AS (
         SELECT * FROM {{ ref('google_analytics_events_raw') }} 
-        WHERE event_name = 'ta_first_open' 
+        WHERE event_name = 'our_first_open' 
           AND {{ ta_firebase.analyticsDateFilterFor('event_date',extend = 2) }}
           AND event_ts BETWEEN install_ts AND TIMESTAMP_ADD(install_ts, INTERVAL 1 DAY)
         QUALIFY ROW_NUMBER() OVER (PARTITION BY user_pseudo_id ORDER BY event_ts) = 1
@@ -45,6 +45,19 @@ WITH  custom_install_event AS (
           AND event_ts BETWEEN install_ts AND TIMESTAMP_ADD(install_ts, INTERVAL 1 DAY)
         QUALIFY ROW_NUMBER() OVER (PARTITION BY user_pseudo_id ORDER BY event_ts) = 1
 )
+, first_user_session_span AS (
+    SELECT
+        user_pseudo_id,
+        TIMESTAMP_DIFF(
+            MAX(event_ts),
+            MIN(event_ts),
+            SECOND
+        ) AS seconds_in_app_on_install_day
+    FROM {{ ref('google_analytics_events_raw') }}
+    WHERE {{ ta_firebase.analyticsDateFilterFor('event_date', extend = 2) }}
+      AND event_ts BETWEEN install_ts AND TIMESTAMP_ADD(install_ts, INTERVAL 1 DAY)
+    GROUP BY user_pseudo_id
+)
 , user_pseudo_id_to_user_id AS (
         SELECT user_pseudo_id, user_id
         FROM {{ ref('google_analytics_events_raw') }} WHERE user_id IS NOT NULL AND {{ ta_firebase.analyticsDateFilterFor('event_date', extend = 2) }}
@@ -65,15 +78,22 @@ WITH  custom_install_event AS (
 
 
 , data as (
-    SELECT   {{ columnsInSelect | join("\n           , ") }} 
+    SELECT   {{ columnsInSelect | join("\n           , ") }}
+             , custom.user_pseudo_id IS NOT NULL        AS user_pseudo_id_custom_event
+             , ta.user_pseudo_id IS NOT NULL            AS user_pseudo_id_our_first_open
+             , ga.user_pseudo_id IS NOT NULL            AS user_pseudo_id_first_open
+             , anyFirstEvent.user_pseudo_id IS NOT NULL AS user_pseudo_id_any_first_event
+             , span.seconds_in_app_on_install_day
+             , span.seconds_in_app_on_install_day >= {{ var("TA:QUALIFIED_INSTALL_MIN_SECONDS", 10) }} AS has_post_install_events_after_10s
     FROM any_first_event as anyFirstEvent
     FULL OUTER JOIN ga_install_event as ga ON anyFirstEvent.user_pseudo_id = ga.user_pseudo_id
     FULL OUTER JOIN ta_install_event as ta ON anyFirstEvent.user_pseudo_id = ta.user_pseudo_id
     FULL OUTER JOIN custom_install_event as custom ON anyFirstEvent.user_pseudo_id = custom.user_pseudo_id
     LEFT JOIN user_pseudo_id_to_user_id as users ON COALESCE(anyFirstEvent.user_pseudo_id, ga.user_pseudo_id, ta.user_pseudo_id, custom.user_pseudo_id) = users.user_pseudo_id
+    LEFT JOIN first_user_session_span as span    ON COALESCE(anyFirstEvent.user_pseudo_id, ga.user_pseudo_id, ta.user_pseudo_id, custom.user_pseudo_id) = span.user_pseudo_id
     WHERE True 
 )
--- SELECT  COUNT(1) , COUNT(DISTINCT(user_pseudo_id))
+-- SELECT  COUNT(1) , COUNT(DISTINCT(user_pseudo_id)) 
 SELECT *
 FROM data 
 WHERE {{ ta_firebase.analyticsDateFilterFor('event_date') }}