diff --git a/cumulus_etl/etl/tasks/base.py b/cumulus_etl/etl/tasks/base.py
index d6b2163..5091fc5 100644
--- a/cumulus_etl/etl/tasks/base.py
+++ b/cumulus_etl/etl/tasks/base.py
@@ -326,7 +326,13 @@ def is_unique(row):
             id_set.add(row_id)
             return True
 
-        return [row for row in rows if is_unique(row)]
+        # Uniquify in reverse, so that later rows will be preferred.
+        # This makes it easy to throw updates of source data alongside the originals,
+        # by either appending to an ndjson file or adding new files that sort later.
+        rows = [row for row in reversed(rows) if is_unique(row)]
+        # But keep original row ordering for cleanliness of ndjson output-format ordering.
+        rows.reverse()
+        return rows
 
     def _write_one_table_batch(self, rows: list[dict], table_index: int, batch_index: int) -> bool:
         # Checkpoint scrubber data before writing to the store, because if we get interrupted, it's safer to have an
diff --git a/tests/etl/test_etl_cli.py b/tests/etl/test_etl_cli.py
index 033d1b8..f388d50 100644
--- a/tests/etl/test_etl_cli.py
+++ b/tests/etl/test_etl_cli.py
@@ -455,14 +455,26 @@ async def test_etl_job_deltalake(self):
             {
                 "_delta_log/00000000000000000000.json",  # create
                 "_delta_log/.00000000000000000000.json.crc",
+                "_delta_log/00000000000000000000.crc",
+                "_delta_log/.00000000000000000000.crc.crc",
                 "_delta_log/00000000000000000001.json",  # merge
                 "_delta_log/.00000000000000000001.json.crc",
+                "_delta_log/00000000000000000001.crc",
+                "_delta_log/.00000000000000000001.crc.crc",
                 "_delta_log/00000000000000000002.json",  # optimize
                 "_delta_log/.00000000000000000002.json.crc",
+                "_delta_log/00000000000000000002.crc",
+                "_delta_log/.00000000000000000002.crc.crc",
                 "_delta_log/00000000000000000003.json",  # vacuum start
                 "_delta_log/.00000000000000000003.json.crc",
+                "_delta_log/00000000000000000003.crc",
+                "_delta_log/.00000000000000000003.crc.crc",
                 "_delta_log/00000000000000000004.json",  # vacuum end
                 "_delta_log/.00000000000000000004.json.crc",
+                "_delta_log/00000000000000000004.crc",
+                "_delta_log/.00000000000000000004.crc.crc",
+                "_delta_log/_last_vacuum_info",
+                "_delta_log/._last_vacuum_info.crc",
                 "_symlink_format_manifest/manifest",
                 "_symlink_format_manifest/.manifest.crc",
             },
diff --git a/tests/etl/test_tasks.py b/tests/etl/test_tasks.py
index 03ea31e..55fdb27 100644
--- a/tests/etl/test_tasks.py
+++ b/tests/etl/test_tasks.py
@@ -80,19 +80,28 @@ def test_task_selection_ordering(self, user_tasks, expected_tasks):
     async def test_drop_duplicates(self):
         """Verify that we run() will drop duplicate rows inside an input batch."""
         # Two "A" ids and one "B" id
-        self.make_json("Patient", "A")
-        self.make_json("Patient", "A")
+        self.make_json("Patient", "A", birthDate="2020")
+        self.make_json("Patient", "A", birthDate="2021")
         self.make_json("Patient", "B")
 
         await basic_tasks.PatientTask(self.job_config, self.scrubber).run()
 
-        # Confirm that only one version of patient A got stored
+        # Confirm that only the later version of patient A got stored
         self.assertEqual(1, self.format.write_records.call_count)
         batch = self.format.write_records.call_args[0][0]
-        self.assertEqual(2, len(batch.rows))
         self.assertEqual(
-            {self.codebook.db.patient("A"), self.codebook.db.patient("B")},
-            {row["id"] for row in batch.rows},
+            batch.rows,
+            [  # Output ordering is guaranteed to be stable
+                {
+                    "resourceType": "Patient",
+                    "id": self.codebook.db.patient("A"),
+                    "birthDate": "2021",  # the row that came later won
+                },
+                {
+                    "resourceType": "Patient",
+                    "id": self.codebook.db.patient("B"),
+                },
+            ],
         )
 
     async def test_batch_write_errors_saved(self):