diff --git a/README.md b/README.md
index 5cd33b32..3f3a9c81 100644
--- a/README.md
+++ b/README.md
@@ -110,7 +110,7 @@ docs = ann.get_docs(limit=10, offset=0, order_by='x', ascending=True)
 After you have indexed the `docs`, you can update the docs in the index by calling `ann.update()`:
 
 ```python
-updated_docs = [{'id': '0', 'embedding': [], 'price': 6}]
+updated_docs = [{'id': '0', 'embedding': np.random.random([128]).astype(np.float32), 'price': 6}]
 
 ann.update(updated_docs)
 ```
diff --git a/annlite/index.py b/annlite/index.py
index 22dd8b36..d642bcfe 100644
--- a/annlite/index.py
+++ b/annlite/index.py
@@ -56,6 +56,7 @@ def __init__(
         self,
         n_dim: int,
         metric: Union[str, Metric] = 'cosine',
+        embedding_field: str = 'embedding',
         n_cells: int = 1,
         n_subvectors: Optional[int] = None,
         n_clusters: Optional[int] = 256,
@@ -90,6 +91,7 @@ def __init__(
         self.n_probe = max(n_probe, n_cells)
         self.n_cells = n_cells
         self.size_limit = 2048
+        self._embedding_field = embedding_field
 
         if isinstance(metric, str):
             metric = Metric.from_string(metric)
@@ -172,7 +174,7 @@ def __init__(
             total_size = 0
             # TODO: add a progress bar
             for docs in self.documents_generator(0, batch_size=1024):
-                x = np.array([doc['embedding'] for doc in docs])
+                x = np.array([doc[self._embedding_field] for doc in docs])
                 total_size += x.shape[0]
                 self.partial_train(x, auto_save=True, force_train=True)
                 if total_size >= MAX_TRAINING_DATA_SIZE:
@@ -280,8 +282,7 @@ def index(self, docs: 'List', **kwargs):
         if not self.is_trained:
             raise RuntimeError(f'The indexer is not trained, cannot add new documents')
 
-        # TODO: Obtain the embeddings from the dict or change index signature
-        x = np.array([doc['embedding'] for doc in docs])
+        x = np.array([doc[self._embedding_field] for doc in docs])
         n_data, _ = self._sanity_check(x)
 
         assigned_cells = (
@@ -312,7 +313,7 @@ def update(
             raise RuntimeError(f'The indexer is not trained, cannot add new documents')
 
         # TODO: Obtain the embeddings from the dict or change index signature
-        x = np.array([doc['embedding'] for doc in docs])
+        x = np.array([doc[self._embedding_field] for doc in docs])
         n_data, _ = self._sanity_check(x)
 
         assigned_cells = (
@@ -347,7 +348,7 @@ def search(
         if not self.is_trained:
             raise RuntimeError(f'The indexer is not trained, cannot add new documents')
 
-        query_np = np.array([doc['embedding'] for doc in docs])
+        query_np = np.array([doc[self._embedding_field] for doc in docs])
 
         _, match_docs = self.search_by_vectors(
             query_np, filter=filter, limit=limit, include_metadata=include_metadata
@@ -778,7 +779,7 @@ def _rebuild_index_from_local(self):
                     f'Rebuild the index of cell-{cell_id} ({cell_size} docs)...'
                 )
                 for docs in self.documents_generator(cell_id, batch_size=10240):
-                    x = np.array([doc['embedding'] for doc in docs])
+                    x = np.array([doc[self._embedding_field] for doc in docs])
 
                     assigned_cells = np.ones(len(docs), dtype=np.int64) * cell_id
                     super().insert(x, assigned_cells, docs, only_index=True)
diff --git a/tests/test_index.py b/tests/test_index.py
index 1ef0e3f8..9ded2524 100644
--- a/tests/test_index.py
+++ b/tests/test_index.py
@@ -268,7 +268,6 @@ def test_local_backup_restore(tmpdir):
     index = AnnLite(n_dim=D, data_path=tmpdir / 'workspace' / '0')
     index.index(docs)
 
-    tmpname = uuid.uuid4().hex
     index.backup()
     index.close()
 
@@ -278,3 +277,24 @@ def test_local_backup_restore(tmpdir):
     status = index.stat
     assert int(status['total_docs']) == N
     assert int(status['index_size']) == N
+
+
+def test_index_search_different_field(tmpdir):
+    X = np.random.random((N, D)).astype(
+        np.float32
+    )  # 10,000 128-dim vectors to be indexed
+
+    index = AnnLite(
+        n_dim=D, data_path=str(tmpdir), embedding_field='encoding', metric='euclidean'
+    )
+    docs = [dict(id=f'{i}', encoding=X[i]) for i in range(N)]
+    index.index(docs)
+    query = [dict(encoding=X[i]) for i in range(5)]
+
+    matches = index.search(query)
+
+    for i in range(len(matches[0]) - 1):
+        assert (
+            matches[0][i]['scores']['euclidean']
+            <= matches[0][i + 1]['scores']['euclidean']
+        )