diff --git a/smart_kit/start_points/main_loop_async_http.py b/smart_kit/start_points/main_loop_async_http.py
index 0d92ed50..915fbceb 100644
--- a/smart_kit/start_points/main_loop_async_http.py
+++ b/smart_kit/start_points/main_loop_async_http.py
@@ -127,15 +127,18 @@ async def handle_message(self, message: SmartAppFromMessage) -> typing.Tuple[int
             code = 204
             log(f"OUTGOING DATA: {answer.masked_value} with code: {code}",
                 params={log_const.KEY_NAME: "outgoing_policy_message"}, user=user)
+            monitoring.counter_outgoing(self.app_name, answer.command.name, answer.command, user)
             return code, "NO CONTENT", answer
 
         answer_message = SmartAppToMessage(
-            answer, message, request=None, validators=self.to_msg_validators, masking_fields=self.masking_fields)
+            answer, message, request=None, validators=self.to_msg_validators, masking_fields=self.masking_fields
+        )
         if answer_message.validate():
             code = 200
             log_answer = str(answer_message.masked_value).replace("%", "%%")
             log(f"OUTGOING DATA: {log_answer} with code: {code}",
                 params={log_const.KEY_NAME: "outgoing_policy_message"}, user=user)
+            monitoring.counter_outgoing(self.app_name, answer.command.name, answer.command, user)
             return code, "OK", answer_message
         else:
             code = 500
@@ -143,6 +146,7 @@ async def handle_message(self, message: SmartAppFromMessage) -> typing.Tuple[int
                 self.BAD_ANSWER_COMMAND, message=message, request=None, masking_fields=self.masking_fields)
             log(f"OUTGOING DATA: {answer.masked_value} with code: {code}",
                 params={log_const.KEY_NAME: "outgoing_policy_message"}, user=user)
+            monitoring.counter_outgoing(self.app_name, answer.command.name, answer.command, user)
             return code, "BAD ANSWER", answer
 
     async def process_message(self, message: SmartAppFromMessage, *args, **kwargs):
@@ -153,6 +157,7 @@ async def process_message(self, message: SmartAppFromMessage, *args, **kwargs):
 
         with StatsTimer() as load_timer:
             user = await self.load_user(db_uid, message)
+        monitoring.sampling_load_time(self.app_name, load_timer.secs)
         stats += "Loading time: {} msecs\n".format(load_timer.msecs)
         with StatsTimer() as script_timer:
             commands = await self.model.answer(message, user)
@@ -164,6 +169,7 @@ async def process_message(self, message: SmartAppFromMessage, *args, **kwargs):
         stats += "Script time: {} msecs\n".format(script_timer.msecs)
         with StatsTimer() as save_timer:
             await self.save_user(db_uid, user, message)
+        monitoring.sampling_save_time(self.app_name, save_timer.secs)
         stats += "Saving time: {} msecs\n".format(save_timer.msecs)
         log(stats, params={log_const.KEY_NAME: "timings"})
         await self.postprocessor.postprocess(user, message)
diff --git a/smart_kit/start_points/main_loop_http.py b/smart_kit/start_points/main_loop_http.py
index 6d677b83..76a319b3 100644
--- a/smart_kit/start_points/main_loop_http.py
+++ b/smart_kit/start_points/main_loop_http.py
@@ -9,6 +9,7 @@
 from core.configs.global_constants import CALLBACK_ID_HEADER
 from core.logging.logger_utils import log
 from core.message.from_message import SmartAppFromMessage, basic_error_message
+from core.monitoring.monitoring import monitoring
 from core.utils.stats_timer import StatsTimer
 from smart_kit.compatibility.commands import combine_commands
 from smart_kit.message.smartapp_to_message import SmartAppToMessage
@@ -70,6 +71,7 @@ def process_message(self, message: SmartAppFromMessage, *args, **kwargs):
         db_uid = message.db_uid
         with StatsTimer() as load_timer:
             user = self.loop.run_until_complete(self.load_user(db_uid, message))
+        monitoring.sampling_load_time(self.app_name, load_timer.secs)
         stats += "Loading time: {} msecs\n".format(load_timer.msecs)
         with StatsTimer() as script_timer:
             commands = asyncio.get_event_loop().run_until_complete(self.model.answer(message, user))
@@ -81,6 +83,7 @@ def process_message(self, message: SmartAppFromMessage, *args, **kwargs):
         stats += "Script time: {} msecs\n".format(script_timer.msecs)
         with StatsTimer() as save_timer:
             self.loop.run_until_complete(self.save_user(db_uid, user, message))
+        monitoring.sampling_save_time(self.app_name, save_timer.secs)
         stats += "Saving time: {} msecs\n".format(save_timer.msecs)
         log(stats, user=user, params={log_const.KEY_NAME: "timings"})
         self.loop.run_until_complete(self.postprocessor.postprocess(user, message))
diff --git a/smart_kit/start_points/main_loop_kafka.py b/smart_kit/start_points/main_loop_kafka.py
index 0d3cf87b..44d1bafe 100644
--- a/smart_kit/start_points/main_loop_kafka.py
+++ b/smart_kit/start_points/main_loop_kafka.py
@@ -356,8 +356,6 @@ def _generate_answers(self, user, commands, message, **kwargs):
             else:
                 answers.append(SmartAppToMessage(self.BAD_ANSWER_COMMAND, message=message, request=request))
 
-            monitoring.counter_outgoing(self.app_name, command.name, answer, user)
-
         return answers
 
     def _get_timeout_from_message(self, orig_message_raw: Dict, callback_id, headers):
@@ -603,6 +601,7 @@ async def _send_request(self, user: BaseUser, answer: SmartAppToMessage, mq_mess
         request_params["mq_message"] = mq_message
         request_params["payload"] = answer.value
         request_params["masked_value"] = answer.masked_value
+        monitoring.counter_outgoing(self.app_name, answer.command.name, answer.command, user)
         await request.run(answer.value.encode(), request_params)
         self._log_request(user, request, answer, mq_message)