diff --git a/integration_tests/models/plugins/spark/spark_external.yml b/integration_tests/models/plugins/spark/spark_external.yml
index bd2af2df..82a2af80 100644
--- a/integration_tests/models/plugins/spark/spark_external.yml
+++ b/integration_tests/models/plugins/spark/spark_external.yml
@@ -41,6 +41,12 @@ sources:
         columns: *cols-of-the-people
         tests: *equal-to-the-people
 
+      - name: people_csv_partitioned_inferred_using
+        external:
+          <<: *csv-people-using
+          recover_partitions: true
+        tests: *equal-to-the-people
+
 # ----- TODO: hive format
         
 #      - name: people_csv_unpartitioned_hive_format
diff --git a/macros/plugins/spark/helpers/recover_partitions.sql b/macros/plugins/spark/helpers/recover_partitions.sql
index 2d20212b..a79de3e2 100644
--- a/macros/plugins/spark/helpers/recover_partitions.sql
+++ b/macros/plugins/spark/helpers/recover_partitions.sql
@@ -2,7 +2,7 @@
     {# https://docs.databricks.com/sql/language-manual/sql-ref-syntax-ddl-alter-table.html #}
 
     {% set ddl %}
-    {%- if source_node.external.partitions and source_node.external.using and source_node.external.using|lower != 'delta' -%}
+    {%- if (source_node.external.partitions or source_node.external.recover_partitions) and source_node.external.using and source_node.external.using|lower != 'delta' -%}
         ALTER TABLE {{ source(source_node.source_name, source_node.name) }} RECOVER PARTITIONS
     {%- endif -%}
     {% endset %}
diff --git a/sample_sources/spark.yml b/sample_sources/spark.yml
index 658e198c..5e49b6fa 100644
--- a/sample_sources/spark.yml
+++ b/sample_sources/spark.yml
@@ -30,3 +30,15 @@ sources:
           - name: contexts
             data_type: string
             description: "Contexts attached to event by Tracker"
+
+- name: event_inferred_schema
+        description: "Snowplow events stored as partitioned parquet files in HDFS with inferred schema"
+        external:
+          # File path can contain partitions such as: hdfs://.../events/my_partition=2022-03-01/events1.parquet
+          # These partitions are excluded from 'location'.
+          location: 'hdfs://.../events/'
+          using: parquet
+
+          # Setting recover_partitions to true causes partitions to be refreshed,
+          # even though partitions are not explicitly specified.
+          recover_partitions: true