diff --git a/arroyo/backends/kafka/configuration.py b/arroyo/backends/kafka/configuration.py
index fe984287..20cfb9e6 100644
--- a/arroyo/backends/kafka/configuration.py
+++ b/arroyo/backends/kafka/configuration.py
@@ -237,7 +237,7 @@ def build_kafka_consumer_configuration(
     bootstrap_servers: Optional[Sequence[str]] = None,
     override_params: Optional[Mapping[str, Any]] = None,
     strict_offset_reset: Optional[bool] = None,
-    enable_auto_commit: bool = False,
+    enable_auto_commit: bool = True,
 ) -> KafkaBrokerConfig:
 
     if auto_offset_reset is None:
diff --git a/docs/source/what_for.rst b/docs/source/what_for.rst
index 7e2d04fd..c7d721f6 100644
--- a/docs/source/what_for.rst
+++ b/docs/source/what_for.rst
@@ -79,11 +79,15 @@ reached its destination in the following way:
 
    # add this value to the config:
    "enable.auto.commit": "false"
+   "enable.auto.offset.store": "false"
    # -------
    message = consumer.poll(timeout=0)
    send_to_destination(process_message(message))
    consumer.commit(message.offset())
 
+Note: Arroyo still uses rdkafka's autocommit, but only to commit stored
+offsets. Offsets are still stored explicitly rather than on poll.
+
 High Throughput
 ---------------