diff --git a/src/eodash_catalog/utils.py b/src/eodash_catalog/utils.py
index caa04847..f7411869 100644
--- a/src/eodash_catalog/utils.py
+++ b/src/eodash_catalog/utils.py
@@ -636,7 +636,7 @@ def save_items(
         output_path = f"{buildcatpath}/{colpath}"
         os.makedirs(output_path, exist_ok=True)
         parquet_path = f"{output_path}/items.parquet"
-        stacgp.arrow.to_parquet(table, parquet_path)
+        stacgp.arrow.to_parquet(table, parquet_path, write_page_index=True)
         extents = extract_extent_from_geoparquet(table)
         collection.extent.temporal = extents[0]
         collection.extent.spatial = extents[1]
diff --git a/tests/test_geoparquet.py b/tests/test_geoparquet.py
index 9ef5d69a..04136f5e 100644
--- a/tests/test_geoparquet.py
+++ b/tests/test_geoparquet.py
@@ -58,6 +58,11 @@ def test_geoparquet_geojson_items(catalog_output_folder):
         )
         assert parquet_asset["file:size"] == os.path.getsize(items_path)
 
+    # page index lets the client skip pages instead of reading whole column chunks
+    column = pa.parquet.ParquetFile(items_path).metadata.row_group(0).column(0)
+    assert column.has_column_index
+    assert column.has_offset_index
+
     with open(items_path, "rb") as fp:
         table = pa.parquet.read_table(fp)
         items = list(stac_gp.arrow.stac_table_to_items(table))