feat(web-analytics): Add bugfix for bounce rate workaround (#22827)

* Add bugfix for bounce rate workaround

* Parametrize the bounce rate test across different pageview modes

* Add some more comments

posthog/hogql/database/schema/sessions.py

@@ -205,6 +205,7 @@ def arg_max_merge_field(field_name: str) -> ast.Call:
     aggregate_fields["$is_bounce"] = ast.Call(
         name="if",
         args=[
+            # if pageview_count is 0, return NULL so it doesn't contribute towards the bounce rate either way
             ast.Call(name="equals", args=[bounce_pageview_count, ast.Constant(value=0)]),
             ast.Constant(value=None),
             ast.Call(
@@ -213,10 +214,13 @@ def arg_max_merge_field(field_name: str) -> ast.Call:
                     ast.Call(
                         name="or",
                         args=[
-                            ast.Call(name="greater", args=[aggregate_fields["$pageview_count"], ast.Constant(value=1)]),
+                            # if > 1 pageview, not a bounce
+                            ast.Call(name="greater", args=[bounce_pageview_count, ast.Constant(value=1)]),
+                            # if > 0 autocapture events, not a bounce
                             ast.Call(
                                 name="greater", args=[aggregate_fields["$autocapture_count"], ast.Constant(value=0)]
                             ),
+                            # if session duration >= 10 seconds, not a bounce
                             ast.Call(
                                 name="greaterOrEquals",
                                 args=[aggregate_fields["$session_duration"], ast.Constant(value=10)],

posthog/hogql/database/schema/test/test_sessions.py

@@ -1,8 +1,11 @@
+from parameterized import parameterized
+
 from posthog.hogql import ast
 from posthog.hogql.database.schema.sessions import get_lazy_session_table_properties
 from posthog.hogql.parser import parse_select
 from posthog.hogql.query import execute_hogql_query
 from posthog.models.property_definition import PropertyType
+from posthog.schema import HogQLQueryModifiers, BounceRatePageViewMode
 from posthog.test.base import (
     APIBaseTest,
     ClickhouseTestMixin,
@@ -139,87 +142,89 @@ def test_persons_and_sessions_on_events(self):
         self.assertEqual(row1, (p1.uuid, "source1"))
         self.assertEqual(row2, (p2.uuid, "source2"))
 
-    def test_bounce_rate(self):
+    @parameterized.expand([(BounceRatePageViewMode.uniq_urls,), (BounceRatePageViewMode.count_pageviews,)])
+    def test_bounce_rate(self, bounceRatePageViewMode):
         # person with 2 different sessions
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d1",
-            properties={"$session_id": "s1a"},
+            properties={"$session_id": "s1a", "$current_url": "https://example.com/1"},
             timestamp="2023-12-02",
         )
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d1",
-            properties={"$session_id": "s1a"},
+            properties={"$session_id": "s1a", "$current_url": "https://example.com/2"},
             timestamp="2023-12-03",
         )
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d1",
-            properties={"$session_id": "s1b"},
+            properties={"$session_id": "s1b", "$current_url": "https://example.com/3"},
             timestamp="2023-12-12",
         )
         # session with 1 pageview
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d2",
-            properties={"$session_id": "s2"},
+            properties={"$session_id": "s2", "$current_url": "https://example.com/4"},
             timestamp="2023-12-11",
         )
         # session with 1 pageview and 1 autocapture
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d3",
-            properties={"$session_id": "s3"},
+            properties={"$session_id": "s3", "$current_url": "https://example.com/5"},
             timestamp="2023-12-11",
         )
         _create_event(
             event="$autocapture",
             team=self.team,
             distinct_id="d3",
-            properties={"$session_id": "s3"},
+            properties={"$session_id": "s3", "$current_url": "https://example.com/5"},
             timestamp="2023-12-11",
         )
         # short session with a pageleave
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d4",
-            properties={"$session_id": "s4"},
+            properties={"$session_id": "s4", "$current_url": "https://example.com/6"},
             timestamp="2023-12-11T12:00:00",
         )
         _create_event(
             event="$pageleave",
             team=self.team,
             distinct_id="d4",
-            properties={"$session_id": "s4"},
+            properties={"$session_id": "s4", "$current_url": "https://example.com/6"},
             timestamp="2023-12-11T12:00:01",
         )
         # long session with a pageleave
         _create_event(
             event="$pageview",
             team=self.team,
             distinct_id="d5",
-            properties={"$session_id": "s5"},
+            properties={"$session_id": "s5", "$current_url": "https://example.com/7"},
             timestamp="2023-12-11T12:00:00",
         )
         _create_event(
             event="$pageleave",
             team=self.team,
             distinct_id="d5",
-            properties={"$session_id": "s5"},
+            properties={"$session_id": "s5", "$current_url": "https://example.com/7"},
             timestamp="2023-12-11T12:00:11",
         )
         response = execute_hogql_query(
             parse_select(
                 "select $is_bounce, session_id from sessions ORDER BY session_id",
             ),
             self.team,
+            modifiers=HogQLQueryModifiers(bounceRatePageViewMode=bounceRatePageViewMode),
         )
         self.assertEqual(
             [