diff --git a/rust_snuba/src/config.rs b/rust_snuba/src/config.rs
index c40f64e9ef..ff05473a65 100644
--- a/rust_snuba/src/config.rs
+++ b/rust_snuba/src/config.rs
@@ -105,6 +105,7 @@ pub struct EnvConfig {
     pub default_retention_days: u16,
     pub lower_retention_days: u16,
     pub valid_retention_days: HashSet<u16>,
+    pub record_cogs: bool,
 }
 
 impl Default for EnvConfig {
@@ -116,6 +117,7 @@ impl Default for EnvConfig {
             default_retention_days: 90,
             lower_retention_days: 30,
             valid_retention_days: [30, 90].iter().cloned().collect(),
+            record_cogs: false,
         }
     }
 }
diff --git a/rust_snuba/src/factory.rs b/rust_snuba/src/factory.rs
index 07f5994613..ebe432af20 100644
--- a/rust_snuba/src/factory.rs
+++ b/rust_snuba/src/factory.rs
@@ -131,10 +131,10 @@ impl ProcessingStrategyFactory<KafkaPayload> for ConsumerStrategyFactory {
 
         let cogs_label = get_cogs_label(&self.storage_config.message_processor.python_class_name);
 
-        // Produce cogs if generic metrics and we are not skipping writes
+        // Produce cogs if generic metrics AND we are not skipping writes AND record_cogs is true
         let next_step: Box<dyn ProcessingStrategy<BytesInsertBatch>> =
-            match (self.skip_write, cogs_label) {
-                (false, Some(resource_id)) => Box::new(RecordCogs::new(
+            match (self.skip_write, self.env_config.record_cogs, cogs_label) {
+                (false, true, Some(resource_id)) => Box::new(RecordCogs::new(
                     next_step,
                     resource_id,
                     self.accountant_topic_config.broker_config.clone(),
diff --git a/snuba/consumers/consumer_config.py b/snuba/consumers/consumer_config.py
index 406edb28d5..3cc083ee5a 100644
--- a/snuba/consumers/consumer_config.py
+++ b/snuba/consumers/consumer_config.py
@@ -50,6 +50,7 @@ class EnvConfig:
     default_retention_days: int
     lower_retention_days: int
     valid_retention_days: list[int]
+    record_cogs: bool
 
 
 @dataclass(frozen=True)
@@ -124,6 +125,7 @@ def _resolve_env_config() -> EnvConfig:
     default_retention_days = settings.DEFAULT_RETENTION_DAYS
     lower_retention_days = settings.LOWER_RETENTION_DAYS
     valid_retention_days = list(settings.VALID_RETENTION_DAYS)
+    record_cogs = settings.RECORD_COGS
     return EnvConfig(
         sentry_dsn=sentry_dsn,
         dogstatsd_host=dogstatsd_host,
@@ -131,6 +133,7 @@ def _resolve_env_config() -> EnvConfig:
         default_retention_days=default_retention_days,
         lower_retention_days=lower_retention_days,
         valid_retention_days=valid_retention_days,
+        record_cogs=record_cogs,
     )
 
 
diff --git a/snuba/datasets/processors/generic_metrics_processor.py b/snuba/datasets/processors/generic_metrics_processor.py
index 4aa2efb739..eb09fdbb92 100644
--- a/snuba/datasets/processors/generic_metrics_processor.py
+++ b/snuba/datasets/processors/generic_metrics_processor.py
@@ -3,7 +3,6 @@
 import zlib
 from abc import ABC, abstractmethod, abstractproperty
 from datetime import datetime, timezone
-from random import random
 from typing import (
     Any,
     Iterable,
@@ -18,6 +17,7 @@
 from sentry_kafka_schemas.schema_types.snuba_generic_metrics_v1 import GenericMetric
 from usageaccountant import UsageUnit
 
+from snuba import settings
 from snuba.cogs.accountant import record_cogs
 from snuba.consumers.types import KafkaMessageMetadata
 from snuba.datasets.events_format import EventTooOld, enforce_retention
@@ -37,7 +37,6 @@
 )
 from snuba.datasets.processors import DatasetMessageProcessor
 from snuba.processor import InsertBatch, ProcessedMessage, _ensure_valid_date
-from snuba.state import get_config
 
 logger = logging.getLogger(__name__)
 
@@ -167,7 +166,7 @@ def process_message(
         )
 
     def __record_cogs(self, message: GenericMetric) -> None:
-        if random() < (get_config("gen_metrics_processor_cogs_probability") or 0):
+        if settings.RECORD_COGS:
             record_cogs(
                 resource_id=self._resource_id,
                 app_feature=f"genericmetrics_{message['use_case_id']}",
diff --git a/snuba/settings/__init__.py b/snuba/settings/__init__.py
index 498a54cf6d..1006523e6e 100644
--- a/snuba/settings/__init__.py
+++ b/snuba/settings/__init__.py
@@ -188,6 +188,9 @@ class RedisClusters(TypedDict):
 # Query Recording Options
 RECORD_QUERIES = False
 
+# Record COGS
+RECORD_COGS = False
+
 # Runtime Config Options
 CONFIG_MEMOIZE_TIMEOUT = 10
 CONFIG_STATE: Mapping[str, Optional[Any]] = {}
diff --git a/tests/datasets/test_generic_metrics_processor.py b/tests/datasets/test_generic_metrics_processor.py
index f98b6c8060..5e76d7a8bc 100644
--- a/tests/datasets/test_generic_metrics_processor.py
+++ b/tests/datasets/test_generic_metrics_processor.py
@@ -6,7 +6,7 @@
 import pytest
 from usageaccountant import UsageUnit
 
-from snuba import state
+from snuba import settings
 from snuba.datasets.metrics_messages import InputType
 from snuba.datasets.processors.generic_metrics_processor import (
     GenericDistributionsMetricsProcessor,
@@ -154,7 +154,8 @@ def test_record_cogs(dis_processor: GenericDistributionsMetricsProcessor) -> Non
         "retention_days": 22,
         "mapping_meta": MAPPING_META_COMMON,
     }
-    state.set_config("gen_metrics_processor_cogs_probability", 1.0)
+    settings.RECORD_COGS = True
+
     with mock.patch(
         "snuba.datasets.processors.generic_metrics_processor.record_cogs"
     ) as record_cogs: