fix(insights): coerce aggregation_target type in multi-source funnels

A funnel that combines series from multiple source tables (e.g. a regular
events step plus a data-warehouse step) builds one subquery per source and
combines them with UNION ALL. Each source resolves aggregation_target to
its own type — person_id is a UUID, a warehouse column is often a string —
and UNION ALL has no common type for UUID and String, failing the whole
query with NO_COMMON_TYPE.

Coerce every branch's aggregation_target to a string when more than one
source query is unioned. Single-source funnels (the common case) are
unchanged, so the person UUID actor id is preserved there.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: sampennington

posthog/hogql_queries/insights/funnels/funnel_event_query.py

@@ -159,6 +159,15 @@ def to_query(self, skip_entity_filter=False, skip_step_filter=False) -> ast.Sele
         if len(queries) == 1:
             return queries[0]
 
+        # Multiple source tables are combined with UNION ALL, which needs a common
+        # column type. Different sources resolve aggregation_target to different
+        # types (e.g. a person_id UUID and a warehouse string column), so coerce
+        # every branch's aggregation_target to a string.
+        for query in queries:
+            for select_expr in query.select:
+                if isinstance(select_expr, ast.Alias) and select_expr.alias == "aggregation_target":
+                    select_expr.expr = ast.Call(name="toString", args=[select_expr.expr])
+
         # Take the field and alias names from the first query. UNION enforces identical column sets
         # across all selects, which makes this reliable.
         aliased_fields = alias_columns_in_select(queries[0].select, self.EVENT_TABLE_ALIAS)

posthog/hogql_queries/insights/funnels/test/__snapshots__/test_funnel_data_warehouse.ambr

@@ -98,7 +98,7 @@
                   e.step_1 AS step_1
            FROM
              (SELECT toTimeZone(e.timestamp, 'UTC') AS timestamp,
-                     if(not(empty(e__override.distinct_id)), e__override.person_id, e.person_id) AS aggregation_target,
+                     toString(if(not(empty(e__override.distinct_id)), e__override.person_id, e.person_id)) AS aggregation_target,
                      e.uuid AS uuid,
                      e.`$session_id` AS `$session_id`,
                      e.`$window_id` AS `$window_id`,
@@ -114,7 +114,7 @@
                  HAVING ifNull(equals(tupleElement(argMax(tuple(person_distinct_id_overrides.is_deleted), person_distinct_id_overrides.version), 1), 0), 0) SETTINGS optimize_aggregation_in_order=1) AS e__override ON equals(e.distinct_id, e__override.distinct_id)
               WHERE and(equals(e.team_id, 99999), and(and(greaterOrEquals(e.timestamp, toDateTime64('explicit_redacted_timestamp', 6, 'UTC')), lessOrEquals(e.timestamp, toDateTime64('explicit_redacted_timestamp', 6, 'UTC'))), in(e.event, tuple('$pageview'))), ifNull(equals(step_0, 1), 0))
               UNION ALL SELECT toTimeZone(e.created, 'UTC') AS timestamp,
-                               accurateCastOrNull(e.user_id, 'UUID') AS aggregation_target,
+                               toString(accurateCastOrNull(e.user_id, 'UUID')) AS aggregation_target,
                                tupleElement(tuple(throwIf(isNull(e.uuid), 'Encountered a null value in posthog_test_test_table_1.uuid, but a non-null value is required. Please ensure this column contains no null values, or add a filter to exclude rows with null values.'), toUUIDOrDefault(toString(e.uuid), reinterpretAsUUID(hex(MD5(concat('posthog_test_test_table_1_0_', ifNull(toString(e.uuid), ''))))))), 2) AS uuid,
                                NULL AS `$session_id`,
                                NULL AS `$window_id`,
@@ -372,7 +372,7 @@
                   e.step_2 AS step_2
            FROM
              (SELECT toTimeZone(e.created_date, 'UTC') AS timestamp,
-                     coalesce(e.converted_opportunity_id, e.id) AS aggregation_target,
+                     toString(coalesce(e.converted_opportunity_id, e.id)) AS aggregation_target,
                      tupleElement(tuple(throwIf(isNull(e.id), 'Encountered a null value in posthog_test_salesforce_lead.id, but a non-null value is required. Please ensure this column contains no null values, or add a filter to exclude rows with null values.'), toUUIDOrDefault(toString(e.id), reinterpretAsUUID(hex(MD5(concat('posthog_test_salesforce_lead_0_', ifNull(toString(e.id), ''))))))), 2) AS uuid,
                      NULL AS `$session_id`,
                      NULL AS `$window_id`,
@@ -382,7 +382,7 @@
               FROM s3('http://host.docker.internal:19000/posthog/test_storage_bucket-posthog.hogql_queries.insights.funnels.funnel_data_warehouse/posthog_test_salesforce_lead/*.csv', 'object_storage_root_user', 'object_storage_root_password', 'CSVWithNames', '`id` String, `created_date` DateTime64(3, \'UTC\'), `converted_opportunity_id` Nullable(String)') AS e
               WHERE and(and(greaterOrEquals(timestamp, toDateTime64('2024-05-01 00:00:00.000000', 6, 'UTC')), lessOrEquals(timestamp, toDateTime64('2024-06-30 23:59:59.999999', 6, 'UTC'))), ifNull(equals(step_0, 1), 0))
               UNION ALL SELECT toTimeZone(e.created_date, 'UTC') AS timestamp,
-                               e.id AS aggregation_target,
+                               toString(e.id) AS aggregation_target,
                                tupleElement(tuple(throwIf(isNull(e.id), 'Encountered a null value in posthog_test_salesforce_opportunity.id, but a non-null value is required. Please ensure this column contains no null values, or add a filter to exclude rows with null values.'), toUUIDOrDefault(toString(e.id), reinterpretAsUUID(hex(MD5(concat('posthog_test_salesforce_opportunity_1_', ifNull(toString(e.id), ''))))))), 2) AS uuid,
                                NULL AS `$session_id`,
                                NULL AS `$window_id`,
@@ -392,7 +392,7 @@
               FROM s3('http://host.docker.internal:19000/posthog/test_storage_bucket-posthog.hogql_queries.insights.funnels.funnel_data_warehouse/posthog_test_salesforce_opportunity/*.csv', 'object_storage_root_user', 'object_storage_root_password', 'CSVWithNames', '`id` String, `close_date` Nullable(Date), `created_date` DateTime64(3, \'UTC\')') AS e
               WHERE and(and(greaterOrEquals(timestamp, toDateTime64('2024-05-01 00:00:00.000000', 6, 'UTC')), lessOrEquals(timestamp, toDateTime64('2024-06-30 23:59:59.999999', 6, 'UTC'))), ifNull(equals(step_1, 1), 0))
               UNION ALL SELECT e.close_date AS timestamp,
-                               e.id AS aggregation_target,
+                               toString(e.id) AS aggregation_target,
                                tupleElement(tuple(throwIf(isNull(e.id), 'Encountered a null value in posthog_test_salesforce_opportunity.id, but a non-null value is required. Please ensure this column contains no null values, or add a filter to exclude rows with null values.'), toUUIDOrDefault(toString(e.id), reinterpretAsUUID(hex(MD5(concat('posthog_test_salesforce_opportunity_2_', ifNull(toString(e.id), ''))))))), 2) AS uuid,
                                NULL AS `$session_id`,
                                NULL AS `$window_id`,

posthog/hogql_queries/insights/funnels/test/test_funnel_data_warehouse.py

@@ -261,6 +261,41 @@ def test_funnels_data_warehouse_and_regular_nodes(self):
             assert results[0]["count"] == 2
             assert results[1]["count"] == 2
 
+    def test_funnels_data_warehouse_and_regular_nodes_string_aggregation_target(self):
+        # A mixed funnel where the warehouse series aggregates by a plain string
+        # column (not cast to UUID) must not fail the UNION ALL with NO_COMMON_TYPE
+        # against the events series' person_id UUID.
+        table_name = self.setup_data_warehouse()
+        with freeze_time("2025-11-07"):
+            _create_person(
+                distinct_ids=["person1"],
+                team_id=self.team.pk,
+                uuid="bc53b62b-7cc4-b3b8-0688-c6ee3dfb8539",
+            )
+            journeys_for(
+                {"person1": [{"event": "$pageview", "timestamp": datetime(2025, 11, 1, 0, 0, 0)}]},
+                self.team,
+                create_people=False,
+            )
+
+            funnels_query = FunnelsQuery(
+                kind="FunnelsQuery",
+                dateRange=DateRange(date_from="2025-11-01"),
+                series=[
+                    EventsNode(event="$pageview"),
+                    FunnelsDataWarehouseNode(
+                        id=table_name,
+                        table_name=table_name,
+                        id_field="uuid",
+                        aggregation_target_field="user_id",
+                        timestamp_field="created",
+                    ),
+                ],
+            )
+
+            response = FunnelsQueryRunner(query=funnels_query, team=self.team, just_summarize=True).calculate()
+            assert response.results is not None
+
     @snapshot_clickhouse_queries
     def test_funnels_data_warehouse_non_uuid_id_column(self):
         table_name = self.setup_data_warehouse()