diff --git a/snuba/web/rpc/v1/resolvers/R_eap_spans/common/aggregation.py b/snuba/web/rpc/v1/resolvers/R_eap_spans/common/aggregation.py
index 9541452423..67d530f725 100644
--- a/snuba/web/rpc/v1/resolvers/R_eap_spans/common/aggregation.py
+++ b/snuba/web/rpc/v1/resolvers/R_eap_spans/common/aggregation.py
@@ -16,7 +16,7 @@
 
 from snuba.query.dsl import CurriedFunctions as cf
 from snuba.query.dsl import Functions as f
-from snuba.query.dsl import column, literal
+from snuba.query.dsl import column
 from snuba.query.expressions import (
     CurriedFunctionCall,
     Expression,
@@ -548,7 +548,9 @@ def get_confidence_interval_column(
                             get_field_existence_expression(aggregation),
                             alias=f"{alias}_N",
                         ),
-                        f.minus(column(f"{alias}_N"), literal(1)),
+                        f.sumIf(
+                            sign_column, get_field_existence_expression(aggregation)
+                        ),
                     ),
                 )
             ),
diff --git a/tests/web/rpc/v1/test_endpoint_time_series/test_endpoint_time_series_extrapolation.py b/tests/web/rpc/v1/test_endpoint_time_series/test_endpoint_time_series_extrapolation.py
index b338a6780a..c71ce8c52e 100644
--- a/tests/web/rpc/v1/test_endpoint_time_series/test_endpoint_time_series_extrapolation.py
+++ b/tests/web/rpc/v1/test_endpoint_time_series/test_endpoint_time_series_extrapolation.py
@@ -474,14 +474,14 @@ def test_sum_unreliable(self) -> None:
 
     def test_avg_unreliable(self) -> None:
         # store a test metric with a value of 1, every second for an hour
-        granularity_secs = 60
+        granularity_secs = 120
         query_duration = 3600
         store_timeseries(
             BASE_TIME,
             1,
             3600,
             # for each time interval we distribute the values from -55 to 64 to keep the avg close to 0
-            metrics=[DummyMetric("test_metric", get_value=lambda x: (x % 60) - 55)],
+            metrics=[DummyMetric("test_metric", get_value=lambda x: (x % 120) - 55)],
             measurements=[
                 DummyMeasurement(
                     "client_sample_rate",
@@ -522,7 +522,7 @@ def test_avg_unreliable(self) -> None:
                 buckets=expected_buckets,
                 data_points=[
                     DataPoint(
-                        data=-25.5,  # (-55 + -54 + ... + -4) / 60 = -25.5
+                        data=4.5,  # (-55 + -54 + ... + 64) / 60 = 4.5
                         data_present=True,
                         reliability=Reliability.RELIABILITY_LOW,
                         avg_sampling_rate=0.0001,