Attempt to shore up the stats table broken down by initial channel type

Both before and after worked fine on my machine, so unclear how this will improve things, but I believe it might
PostHog · Dec 15, 2023 · 2e7f4eb · 2e7f4eb
1 parent 4649e6c
commit 2e7f4eb
Show file tree

Hide file tree

Showing 2 changed files with 85 additions and 5 deletions.
diff --git a/posthog/hogql_queries/web_analytics/stats_table.py b/posthog/hogql_queries/web_analytics/stats_table.py
@@ -104,14 +104,14 @@ def counts_breakdown(self):
 
     (
         match(person.properties.$initial_utm_medium, '^(.*cp.*|ppc|retargeting|paid.*)$') OR
-        properties.$initial_gclid IS NOT NULL OR
-        properties.$initial_gad_source IS NOT NULL
+        person.properties.$initial_gclid IS NOT NULL OR
+        person.properties.$initial_gad_source IS NOT NULL
     ),
     coalesce(
         hogql_lookupPaidSourceType(person.properties.$initial_utm_source),
         hogql_lookupPaidDomainType(person.properties.$initial_referring_domain),
         if(
-            match(properties.$initial_utm_campaign, '^(.*(([^a-df-z]|^)shop|shopping).*)$'),
+            match(person.properties.$initial_utm_campaign, '^(.*(([^a-df-z]|^)shop|shopping).*)$'),
             'Paid Shopping',
             NULL
         ),
@@ -191,6 +191,68 @@ def bounce_breakdown(self):
             case WebStatsBreakdown.Page:
                 # use initial pathname for bounce rate
                 return ast.Call(name="any", args=[ast.Field(chain=["person", "properties", "$initial_pathname"])])
+            case WebStatsBreakdown.InitialChannelType:
+                # use this for now, switch to person.$virt_initial_channel_type when it's working. If fixing or adding
+                # anything to this, keep in sync with channel_type.py
+                return parse_expr(
+                    """
+multiIf(
+    match(any(person.properties.$initial_utm_campaign), 'cross-network'),
+    'Cross Network',
+
+    (
+        match(any(person.properties.$initial_utm_medium), '^(.*cp.*|ppc|retargeting|paid.*)$') OR
+        any(person.properties.$initial_gclid) IS NOT NULL OR
+        any(person.properties.$initial_gad_source) IS NOT NULL
+    ),
+    coalesce(
+        hogql_lookupPaidSourceType(any(person.properties.$initial_utm_source)),
+        hogql_lookupPaidDomainType(any(person.properties.$initial_referring_domain)),
+        if(
+            match(any(person.properties.$initial_utm_campaign), '^(.*(([^a-df-z]|^)shop|shopping).*)$'),
+            'Paid Shopping',
+            NULL
+        ),
+        hogql_lookupPaidMediumType(any(person.properties.$initial_utm_medium)),
+        multiIf (
+            any(person.properties.$initial_gad_source) = '1',
+            'Paid Search',
+
+            match(any(person.properties.$initial_utm_campaign), '^(.*video.*)$'),
+            'Paid Video',
+
+            'Paid Other'
+        )
+    ),
+
+    (
+        any(any(person.properties.$initial_referring_domain)) = '$direct'
+        AND (any(person.properties.$initial_utm_medium) IS NULL OR any(person.properties.$initial_utm_medium) = '')
+        AND (any(person.properties.$initial_utm_source) IS NULL OR any(person.properties.$initial_utm_source) IN ('', '(direct)', 'direct'))
+    ),
+    'Direct',
+
+    coalesce(
+        hogql_lookupOrganicSourceType(any(person.properties.$initial_utm_source)),
+        hogql_lookupOrganicDomainType(any(person.properties.$initial_referring_domain)),
+        if(
+            match(any(person.properties.$initial_utm_campaign), '^(.*(([^a-df-z]|^)shop|shopping).*)$'),
+            'Organic Shopping',
+            NULL
+        ),
+        hogql_lookupOrganicMediumType(any(person.properties.$initial_utm_medium)),
+        multiIf(
+            match(any(person.properties.$initial_utm_campaign), '^(.*video.*)$'),
+            'Organic Video',
+
+            match(any(person.properties.$initial_utm_medium), 'push$'),
+            'Push',
+
+            NULL
+        )
+    )
+)"""
+                )
             case _:
                 return ast.Call(name="any", args=[self.counts_breakdown()])
 

diff --git a/posthog/hogql_queries/web_analytics/test/test_web_stats_table.py b/posthog/hogql_queries/web_analytics/test/test_web_stats_table.py
@@ -35,9 +35,9 @@ def _create_events(self, data, event="$pageview"):
                 )
         return person_result
 
-    def _run_web_stats_table_query(self, date_from, date_to):
+    def _run_web_stats_table_query(self, date_from, date_to, breakdown_by=WebStatsBreakdown.Page):
         query = WebStatsTableQuery(
-            dateRange=DateRange(date_from=date_from, date_to=date_to), properties=[], breakdownBy=WebStatsBreakdown.Page
+            dateRange=DateRange(date_from=date_from, date_to=date_to), properties=[], breakdownBy=breakdown_by
         )
         runner = WebStatsTableQueryRunner(team=self.team, query=query)
         return runner.calculate()
@@ -93,3 +93,21 @@ def test_filter_test_accounts(self):
             [],
             results,
         )
+
+    def test_breakdown_channel_type_doesnt_throw(self):
+        # not really testing the functionality yet, which is tested elsewhere, just that it runs
+        self._create_events(
+            [
+                ("p1", [("2023-12-02", "s1a", "/"), ("2023-12-03", "s1a", "/login"), ("2023-12-13", "s1b", "/docs")]),
+                ("p2", [("2023-12-10", "s2", "/")]),
+            ]
+        )
+
+        results = self._run_web_stats_table_query(
+            "2023-12-01", "2023-12-03", breakdown_by=WebStatsBreakdown.InitialChannelType
+        ).results
+
+        self.assertEqual(
+            [(None, 2, 1, 0)],
+            results,
+        )