diff --git a/CHANGELOG.md b/CHANGELOG.md
index 251b6354..87a74d50 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Changed
+
+- Mimir compactor alert: better failure detection
+
 ### Added
 
 - Add new mimir continuous test alerts:
diff --git a/helm/prometheus-rules/templates/platform/atlas/alerting-rules/mimir.rules.yml b/helm/prometheus-rules/templates/platform/atlas/alerting-rules/mimir.rules.yml
index 30b677d4..c025ce56 100644
--- a/helm/prometheus-rules/templates/platform/atlas/alerting-rules/mimir.rules.yml
+++ b/helm/prometheus-rules/templates/platform/atlas/alerting-rules/mimir.rules.yml
@@ -158,8 +158,24 @@ spec:
         dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
         description: 'Mimir compactor has been failing its compactions for 2 hours.'
         opsrecipe: mimir#mimircompactorfailedcompaction
-      # Query is based on the following upstream mixin alerting rule : https://github.com/grafana/mimir/blob/main/operations/mimir-mixin-compiled/alerts.yaml#L858
-      expr: sum(increase(cortex_compactor_runs_failed_total{reason!="shutdown"}[2h])) by (cluster_id, installation, namespace, pipeline, provider) > 2
+      expr: min by (cluster_id, installation, namespace, provider, pipeline) (time() - (cortex_compactor_last_successful_run_timestamp_seconds > 0 ) ) > 60 * 60 * 2
+      labels:
+        area: platform
+        cancel_if_cluster_status_creating: "true"
+        cancel_if_cluster_status_deleting: "true"
+        cancel_if_cluster_status_updating: "true"
+        cancel_if_outside_working_hours: "true"
+        severity: page
+        team: atlas
+        topic: observability
+    - alert: MimirCompactorFailedCompaction
+      annotations:
+        dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
+        description: 'Mimir compactor has been failing compactions for more than 2 hours since start-up.'
+        opsrecipe: mimir#mimircompactorfailedcompaction
+      # This alert covers the special case at compactor startup, where the "normal" alert would always consider time `0` is more than 2 hours ago, yet we want to let it 2 hours + `for` duration.
+      expr: max(max_over_time(cortex_compactor_last_successful_run_timestamp_seconds{}[2h])) by (cluster_id, installation, namespace, provider, pipeline) == 0
+      for: 2h
       labels:
         area: platform
         cancel_if_cluster_status_creating: "true"
diff --git a/test/tests/providers/capi/capa/platform/atlas/alerting-rules/mimir.rules.test.yml b/test/tests/providers/capi/capa/platform/atlas/alerting-rules/mimir.rules.test.yml
index e25ac35e..80ec37bc 100644
--- a/test/tests/providers/capi/capa/platform/atlas/alerting-rules/mimir.rules.test.yml
+++ b/test/tests/providers/capi/capa/platform/atlas/alerting-rules/mimir.rules.test.yml
@@ -360,13 +360,14 @@ tests:
     input_series:
       - series: 'cortex_compactor_runs_failed_total{reason="error", installation="golem", cluster_id="golem", namespace="mimir", pipeline="testing", provider="capa"}'
         values: "8+0x20 1+0x40 0+0x20 4+0x130 0+0x190"
+      - series: 'cortex_compactor_last_successful_run_timestamp_seconds{installation="golem", cluster_id="golem", namespace="mimir", pipeline="testing", provider="capa"}'
+        # No compactions for 2 hours, then 1 successful one at t+3h, another one at t+4h, then 2 more hours with no successful compaction.
+        values: '0+0x240 14400+0x60 18000x60 21600+0x240'
     alert_rule_test:
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 15m 
-      - alertname: MimirCompactorFailedCompaction
-        eval_time: 55m 
+        eval_time: 60m 
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 120m
+        eval_time: 130m 
         exp_alerts:
           - exp_labels:
               area: platform
@@ -384,12 +385,33 @@ tests:
               topic: observability
             exp_annotations:
               dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
-              description: Mimir compactor has been failing its compactions for 2 hours.
+              description: Mimir compactor has been failing compactions for more than 2 hours since start-up.
               opsrecipe: "mimir#mimircompactorfailedcompaction"
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 205m 
+        eval_time: 250m
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 350m
+        eval_time: 480m 
+      - alertname: MimirCompactorFailedCompaction
+        eval_time: 600m
+        exp_alerts:
+          - exp_labels:
+              area: platform
+              cancel_if_cluster_status_creating: "true"
+              cancel_if_cluster_status_deleting: "true"
+              cancel_if_cluster_status_updating: "true"
+              cancel_if_outside_working_hours: "true"
+              cluster_id: golem
+              installation: "golem"
+              pipeline: "testing"
+              provider: "capa"
+              namespace: mimir
+              severity: page
+              team: atlas
+              topic: observability
+            exp_annotations:
+              dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
+              description: Mimir compactor has been failing its compactions for 2 hours.
+              opsrecipe: "mimir#mimircompactorfailedcompaction"
 
   # Test for MimirContinuousTestFailingOnWrites alert
   - interval: 1m
diff --git a/test/tests/providers/capi/capz/platform/atlas/alerting-rules/mimir.rules.test.yml b/test/tests/providers/capi/capz/platform/atlas/alerting-rules/mimir.rules.test.yml
index d6b37c8e..89d3ca7b 100644
--- a/test/tests/providers/capi/capz/platform/atlas/alerting-rules/mimir.rules.test.yml
+++ b/test/tests/providers/capi/capz/platform/atlas/alerting-rules/mimir.rules.test.yml
@@ -360,13 +360,14 @@ tests:
     input_series:
       - series: 'cortex_compactor_runs_failed_total{reason="error", installation="golem", cluster_id="golem", namespace="mimir", pipeline="testing", provider="capz"}'
         values: "8+0x20 1+0x40 0+0x20 4+0x130 0+0x190"
+      - series: 'cortex_compactor_last_successful_run_timestamp_seconds{installation="golem", cluster_id="golem", namespace="mimir", pipeline="testing", provider="capz"}'
+        # No compactions for 2 hours, then 1 successful one at t+3h, another one at t+4h, then 2 more hours with no successful compaction.
+        values: '0+0x240 14400+0x60 18000x60 21600+0x240'
     alert_rule_test:
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 15m 
-      - alertname: MimirCompactorFailedCompaction
-        eval_time: 55m 
+        eval_time: 60m 
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 120m
+        eval_time: 130m 
         exp_alerts:
           - exp_labels:
               area: platform
@@ -384,12 +385,33 @@ tests:
               topic: observability
             exp_annotations:
               dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
-              description: Mimir compactor has been failing its compactions for 2 hours.
+              description: Mimir compactor has been failing compactions for more than 2 hours since start-up.
               opsrecipe: "mimir#mimircompactorfailedcompaction"
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 205m 
+        eval_time: 250m
       - alertname: MimirCompactorFailedCompaction
-        eval_time: 350m
+        eval_time: 480m 
+      - alertname: MimirCompactorFailedCompaction
+        eval_time: 600m
+        exp_alerts:
+          - exp_labels:
+              area: platform
+              cancel_if_cluster_status_creating: "true"
+              cancel_if_cluster_status_deleting: "true"
+              cancel_if_cluster_status_updating: "true"
+              cancel_if_outside_working_hours: "true"
+              cluster_id: golem
+              installation: "golem"
+              pipeline: "testing"
+              provider: "capz"
+              namespace: mimir
+              severity: page
+              team: atlas
+              topic: observability
+            exp_annotations:
+              dashboard: 09a5c49e9cdb2f2b24c6d184574a07fd/mimir-compactor-resources
+              description: Mimir compactor has been failing its compactions for 2 hours.
+              opsrecipe: "mimir#mimircompactorfailedcompaction"
 
   # Test for MimirContinuousTestFailingOnWrites alert
   - interval: 1m