diff --git a/pkg/models/metrics/metricsrule.go b/pkg/models/metrics/metricsrule.go
index d5c5f2d1f..298869c72 100644
--- a/pkg/models/metrics/metricsrule.go
+++ b/pkg/models/metrics/metricsrule.go
@@ -27,12 +27,16 @@ func MakeWorkloadPromQL(metricName, nsName, resources_filter, wkKind string) str
 		wkKind = DaemonSet
 	case "statefulset":
 		wkKind = StatefulSet
-	default:
-		wkKind = "(.*)"
 	}
 
-	if resources_filter == "" {
-		resources_filter = ".*"
+	if wkKind == "" {
+		resources_filter = Any
+	} else if resources_filter == "" {
+		if strings.Contains(metricName, "pod") {
+			resources_filter = wkKind + ":" + Any
+		} else if strings.Contains(metricName, strings.ToLower(wkKind)) {
+			resources_filter = Any
+		}
 	} else {
 		var prefix string
 
diff --git a/pkg/models/metrics/metricsruleconst.go b/pkg/models/metrics/metricsruleconst.go
index 45f46d77c..ca117f72e 100644
--- a/pkg/models/metrics/metricsruleconst.go
+++ b/pkg/models/metrics/metricsruleconst.go
@@ -76,6 +76,7 @@ const (
 	StatefulSet = "StatefulSet"
 	DaemonSet   = "DaemonSet"
 	Deployment  = "Deployment"
+	Any         = ".*"
 )
 
 const (
@@ -598,6 +599,10 @@ var RulePromQLTmplMap = MetricMap{
 
 	// workload
 	// Join the "container_cpu_usage_seconds_total" metric with "kube_pod_owner" to calculate workload-level resource usage
+	//
+	// Note the name convention:
+	// For hardware resource metrics, combine pod metric name with `workload_`
+	// For k8s resource metrics, must specify the workload type in metric names
 	"workload_pod_cpu_usage":             `round(namespace:workload_cpu_usage:sum{namespace="$2", workload=~"$3"}, 0.001)`,
 	"workload_pod_memory_usage":          `namespace:workload_memory_usage:sum{namespace="$2", workload=~"$3"}`,
 	"workload_pod_memory_usage_wo_cache": `namespace:workload_memory_usage_wo_cache:sum{namespace="$2", workload=~"$3"}`,