diff --git a/src/Storages/Kafka/StorageKafka.cpp b/src/Storages/Kafka/StorageKafka.cpp
index 9c3506742fd..c9f6bcabcc1 100644
--- a/src/Storages/Kafka/StorageKafka.cpp
+++ b/src/Storages/Kafka/StorageKafka.cpp
@@ -783,9 +783,12 @@ void registerStorageKafka(StorageFactory & factory)
 
         if (num_consumers > max_consumers)
         {
-            throw Exception(ErrorCodes::BAD_ARGUMENTS, "Number of consumers can not be bigger than {}, it just doesn't make sense. "
-                            "Note that kafka_num_consumers is not number of consumers for Kafka partitions -- they are managed by Kafka client library. "
-                            "kafka_num_consumers is internal amount of threads for ClickHouse and it shouldn't be big", max_consumers);
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "The number of consumers can not be bigger than {}. "
+                            "A single consumer can read any number of partitions. Extra consumers are relatively expensive, "
+                            "and using a lot of them can lead to high memory and CPU usage. To achieve better performance "
+                            "of getting data from Kafka, consider using a setting kafka_thread_per_consumer=1, "
+                            "and ensure you have enough threads in MessageBrokerSchedulePool (background_message_broker_schedule_pool_size). "
+                            "See also https://clickhouse.com/docs/integrations/kafka/kafka-table-engine#tuning-performance", max_consumers);
         }
         else if (num_consumers < 1)
         {