diff --git a/src/matchbox/server/postgresql/utils/selector.py b/src/matchbox/server/postgresql/utils/selector.py
index ec7cffb..99d213c 100644
--- a/src/matchbox/server/postgresql/utils/selector.py
+++ b/src/matchbox/server/postgresql/utils/selector.py
@@ -3,6 +3,7 @@
 import pyarrow as pa
 from pandas import DataFrame
 from sqlalchemy import Engine, func, select
+from sqlalchemy.dialects.postgresql import array
 from sqlalchemy.orm import Session, aliased
 from sqlalchemy.sql.selectable import Select
 
@@ -17,6 +18,16 @@
 T = TypeVar("T")
 
 
+def hash_to_hex_decode(hash: bytes) -> bytes:
+    """A workround for PostgreSQL so we can compile the query and use ConnectorX."""
+    return func.decode(hash.hex(), "hex")
+
+
+def key_to_sqlalchemy_label(key: str, source: Source) -> str:
+    """Converts a key to the SQLAlchemy LABEL_STYLE_TABLENAME_PLUS_COL."""
+    return f"{source.db_schema}_{source.db_table}_{key}"
+
+
 def get_all_parents(model: Models | list[Models]) -> list[Models]:
     """
     Takes a Models object and returns all items in its parent tree.
@@ -118,12 +129,15 @@ def _tree_to_reachable_stmt(model_tree: list[bytes]) -> Select:
     c1 = aliased(Clusters)
     c2 = aliased(Clusters)
 
+    bytea_array = array([hash_to_hex_decode(m) for m in model_tree])
+    subquery = select(func.unnest(bytea_array).label("hash")).subquery()
+
     dd_stmt = (
         select(DDupeContains.parent, DDupeContains.child)
         .join(c1, DDupeContains.parent == c1.sha1)
         .join(clusters_association, clusters_association.c.child == c1.sha1)
         .join(Models, clusters_association.c.parent == Models.sha1)
-        .where(Models.sha1.in_(model_tree))
+        .where(Models.sha1.in_(select(subquery.c.hash)))
     )
 
     lk_stmt = (
@@ -132,7 +146,7 @@ def _tree_to_reachable_stmt(model_tree: list[bytes]) -> Select:
         .join(c2, LinkContains.child == c2.sha1)
         .join(clusters_association, clusters_association.c.child == c1.sha1)
         .join(Models, clusters_association.c.parent == Models.sha1)
-        .where(Models.sha1.in_(model_tree))
+        .where(Models.sha1.in_(select(subquery.c.hash)))
     )
 
     return dd_stmt.union(lk_stmt)
@@ -164,7 +178,7 @@ def _reachable_to_parent_data_stmt(
         .join(Clusters, Clusters.sha1 == allowed.c.parent)
         .join(clusters_association, clusters_association.c.child == Clusters.sha1)
         .join(Models, clusters_association.c.parent == Models.sha1)
-        .where(Models.sha1 == parent_hash)
+        .where(Models.sha1 == hash_to_hex_decode(parent_hash))
         .cte("root")
     )
 
@@ -236,7 +250,7 @@ def _model_to_hashes(
         the hash key of the SourceData
     """
     parent, child = _parent_to_tree(model, engine=engine)
-    if len(parent) == 0:
+    if not parent:
         raise ValueError(f"Model {model} not found")
     tree = [parent] + child
     reachable_stmt = _tree_to_reachable_stmt(tree)
@@ -316,17 +330,20 @@ def query(
             fields=set([source.db_pk] + fields), pks=mb_hashes["id"].to_pylist()
         )
 
-        # Tablename plus column SQLAlchemy label style
-        right_key = f"{source.db_schema}_{source.db_table}_{source.db_pk}"
-
         joined_table = raw_data.join(
             right_table=mb_hashes,
-            keys=right_key,
+            keys=key_to_sqlalchemy_label(key=source.db_pk, source=source),
             right_keys="id",
             join_type="inner",
         )
 
-        tables.append(joined_table)
+        # Keep only the columns we want
+        keep_cols = ["cluster_hash", "data_hash"] + [
+            key_to_sqlalchemy_label(f, source) for f in fields
+        ]
+        match_cols = [col for col in joined_table.column_names if col in keep_cols]
+
+        tables.append(joined_table.select(match_cols))
 
     result = pa.concat_tables(tables, promote_options="default")
 
diff --git a/test/server/test_adapter.py b/test/server/test_adapter.py
index bf7ed6e..3e89665 100644
--- a/test/server/test_adapter.py
+++ b/test/server/test_adapter.py
@@ -150,7 +150,7 @@ def test_query_with_dedupe_model(
 
     select_crn = selector(
         table=str(crn),
-        fields=["id", "crn"],
+        fields=["company_name", "crn"],
         engine=crn.database.engine,
     )
 
@@ -164,13 +164,13 @@ def test_query_with_dedupe_model(
     assert isinstance(df_crn, DataFrame)
     assert df_crn.shape[0] == 3000
     assert set(df_crn.columns) == {
-        "cluster_sha1",
-        "data_sha1",
+        "cluster_hash",
+        "data_hash",
         "test_crn_crn",
         "test_crn_company_name",
     }
-    assert df_crn.data_sha1.nunique() == 3000
-    assert df_crn.cluster_sha1.nunique() == 1000
+    assert df_crn.data_hash.nunique() == 3000
+    assert df_crn.cluster_hash.nunique() == 1000
 
 
 def test_query_with_link_model():