diff --git a/test_runner/fixtures/benchmark_fixture.py b/test_runner/fixtures/benchmark_fixture.py
index bb8e75902e77..fa3747c08f29 100644
--- a/test_runner/fixtures/benchmark_fixture.py
+++ b/test_runner/fixtures/benchmark_fixture.py
@@ -266,6 +266,16 @@ def record(
         name = f"{self.PROPERTY_PREFIX}_{metric_name}"
         if labels is None:
             labels = {}
+
+        # Sometimes mypy can't catch non-numeric values,
+        # so adding a check here
+        try:
+            float(metric_value)
+        except ValueError as e:
+            raise ValueError(
+                f"`metric_value` (`{metric_value}`) must be a NUMERIC-friendly data type"
+            ) from e
+
         self.property_recorder(
             name,
             {
diff --git a/test_runner/performance/pageserver/test_page_service_batching.py b/test_runner/performance/pageserver/test_page_service_batching.py
index 562094a059d2..2c27368001b3 100644
--- a/test_runner/performance/pageserver/test_page_service_batching.py
+++ b/test_runner/performance/pageserver/test_page_service_batching.py
@@ -116,21 +116,18 @@ def test_throughput(
             # name is not a metric, we just use it to identify the test easily in the `test_...[...]`` notation
         }
     )
-    params.update(
-        {
-            f"pipelining_config.{k}": (v, {})
-            for k, v in dataclasses.asdict(pipelining_config).items()
-        }
-    )
+    # For storing configuration as a metric, insert a fake 0 with labels with actual data
+    params.update({"pipelining_config": (0, {"labels": dataclasses.asdict(pipelining_config)})})
 
     log.info("params: %s", params)
 
     for param, (value, kwargs) in params.items():
         zenbenchmark.record(
             param,
-            metric_value=value,
+            metric_value=float(value),
             unit=kwargs.pop("unit", ""),
             report=MetricReport.TEST_PARAM,
+            labels=kwargs.pop("labels", None),
             **kwargs,
         )