diff --git a/test/perf/run.py b/test/perf/run.py
index 0b286e73..bd203ba2 100644
--- a/test/perf/run.py
+++ b/test/perf/run.py
@@ -24,6 +24,12 @@ from safety import SecretScanner
 import resources
 
 
+# Scenarios run once per invocation. pool waits on the product's real 2 min
+# idle release and 1 min spare cool-down until the scaled cool-down is fixed;
+# its gate is one sample, so it has no noise estimate.
+SINGLE_RUN = {"pool": 1}
+
+
 def summary(values):
     middle = statistics.median(values)
     return {
@@ -150,6 +156,8 @@ def derived(runs):
             "noise": stats["mad"],
             "samples": stats["samples"],
         }
+        if len(values) == 1:
+            metrics[name]["single_sample"] = True
 
     if runs["boot"][0]["checkpoints"]:
         add(
@@ -322,8 +330,9 @@ def main():
     try:
         for name in names:
             raw[name] = []
-            for index in range(args.runs):
-                print(f"{args.mode} {name} {index + 1}/{args.runs}", flush=True)
+            runs = SINGLE_RUN.get(name, args.runs)
+            for index in range(runs):
+                print(f"{args.mode} {name} {index + 1}/{runs}", flush=True)
                 raw[name].append(
                     run_scenario(
                         name,