diff --git a/docs/operator-manual/config-management-plugins.md b/docs/operator-manual/config-management-plugins.md
index c46aa85152c73..4f5601d8533a9 100644
--- a/docs/operator-manual/config-management-plugins.md
+++ b/docs/operator-manual/config-management-plugins.md
@@ -358,8 +358,8 @@ needing to restart Argo CD. You could also easily develop in an in-cluster devel
 locally and your cluster can reach your local machine you could even develop using an ExternalName service.
 
 For large mono-repos sidecars will continue to work where services will take longer to transfer the state to the plugin. Argo CD
-doesn't condone large repos anyway. Because the network traffic between the repo-server and the plugin is over a unix socket it is by
-default more secure.
+doesn't condone large repos anyway. Sidecar plugins (where the repo-server connects over a unix socket) are by
+default more secure than service plugins.
 
 Services allow independent scaling, even using HPAs, installation and upgrading at runtime, and work more like normal services.
 You can add monitoring to both types, but it feels more natural with a service. If you are making your plugin available for public