diff --git a/dlt/destinations/impl/bigquery/bigquery.py b/dlt/destinations/impl/bigquery/bigquery.py
index c3954e0ae5..2b3927e7c9 100644
--- a/dlt/destinations/impl/bigquery/bigquery.py
+++ b/dlt/destinations/impl/bigquery/bigquery.py
@@ -49,6 +49,7 @@
 from dlt.destinations.job_impl import DestinationJsonlLoadJob, DestinationParquetLoadJob
 from dlt.destinations.job_impl import ReferenceFollowupJobRequest
 from dlt.destinations.sql_jobs import SqlMergeFollowupJob
+from dlt.destinations.sql_client import SqlClientBase
 
 
 class BigQueryLoadJob(RunnableLoadJob, HasFollowupJobs):
@@ -142,6 +143,18 @@ def get_job_id_from_file_path(file_path: str) -> str:
 
 
 class BigQueryMergeJob(SqlMergeFollowupJob):
+    @classmethod
+    def _gen_table_setup_clauses(
+        cls, table_chain: Sequence[PreparedTableSchema], sql_client: SqlClientBase[Any]
+    ) -> List[str]:
+        """generate final tables from staging table schema for autodetect tables"""
+        sql: List[str] = []
+        for table in table_chain:
+            if should_autodetect_schema(table):
+                table_name, staging_table_name = sql_client.get_qualified_table_names(table["name"])
+                sql.append(f"CREATE TABLE IF NOT EXISTS {table_name} LIKE {staging_table_name};")
+        return sql
+
     @classmethod
     def gen_key_table_clauses(
         cls,
diff --git a/dlt/destinations/sql_jobs.py b/dlt/destinations/sql_jobs.py
index faa6b50531..2226dfeece 100644
--- a/dlt/destinations/sql_jobs.py
+++ b/dlt/destinations/sql_jobs.py
@@ -167,12 +167,24 @@ def generate_sql(
         merge_strategy = resolve_merge_strategy(
             {root_table["name"]: root_table}, root_table, sql_client.capabilities
         )
+
+        merge_sql = None
         if merge_strategy == "delete-insert":
-            return cls.gen_merge_sql(table_chain, sql_client)
+            merge_sql = cls.gen_merge_sql(table_chain, sql_client)
         elif merge_strategy == "upsert":
-            return cls.gen_upsert_sql(table_chain, sql_client)
+            merge_sql = cls.gen_upsert_sql(table_chain, sql_client)
         elif merge_strategy == "scd2":
-            return cls.gen_scd2_sql(table_chain, sql_client)
+            merge_sql = cls.gen_scd2_sql(table_chain, sql_client)
+
+        # prepend setup code
+        return cls._gen_table_setup_clauses(table_chain, sql_client) + merge_sql
+
+    @classmethod
+    def _gen_table_setup_clauses(
+        cls, table_chain: Sequence[PreparedTableSchema], sql_client: SqlClientBase[Any]
+    ) -> List[str]:
+        """Subclasses may override this method to generate additional sql statements to run before the merge"""
+        return []
 
     @classmethod
     def _gen_key_table_clauses(
@@ -501,12 +513,6 @@ def gen_merge_sql(
             root_table["name"]
         )
 
-        # NOTE: this is bigquery specific code! Move to bigquery merge job
-        # NOTE: we also need to create all child tables
-        # NOTE: this will not work if the schema of the staging table changes in the next run..
-        # in some cases we need to create final tables here
-        sql.append(f"CREATE TABLE IF NOT EXISTS {root_table_name} LIKE {staging_root_table_name};")
-
         # get merge and primary keys from top level
         primary_keys = cls._escape_list(
             get_columns_names_with_prop(root_table, "primary_key"),