diff --git a/CHANGELOG.md b/CHANGELOG.md
index 638e41727f4..39e2dcdc341 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,7 @@
   * `cortex_bucket_stores_gate_queries_concurrent_max`
   * `cortex_bucket_stores_gate_queries_in_flight`
   * `cortex_bucket_stores_gate_duration_seconds`
+* [CHANGE] Metric `cortex_ingester_flush_reasons` has been renamed to `cortex_ingester_series_flushed_total`, and is now incremented during flush, not when series is enqueued for flushing. #2802
 * [FEATURE] Introduced `ruler.for-outage-tolerance`, Max time to tolerate outage for restoring "for" state of alert. #2783
 * [FEATURE] Introduced `ruler.for-grace-period`, Minimum duration between alert and restored "for" state. This is maintained only for alerts with configured "for" time greater than grace period. #2783
 * [FEATURE] Introduced `ruler.resend-delay`, Minimum amount of time to wait before resending an alert to Alertmanager. #2783
diff --git a/pkg/ingester/flush.go b/pkg/ingester/flush.go
index 7dee669454b..5d4a887dea1 100644
--- a/pkg/ingester/flush.go
+++ b/pkg/ingester/flush.go
@@ -135,7 +135,6 @@ func (i *Ingester) sweepSeries(userID string, fp model.Fingerprint, series *memo
 
 	flushQueueIndex := int(uint64(fp) % uint64(i.cfg.ConcurrentFlushes))
 	if i.flushQueues[flushQueueIndex].Enqueue(&flushOp{firstTime, userID, fp, immediate}) {
-		i.metrics.flushReasons.WithLabelValues(flush.String()).Inc()
 		util.Event().Log("msg", "add to flush queue", "userID", userID, "reason", flush, "firstTime", firstTime, "fp", fp, "series", series.metric, "nlabels", len(series.metric), "queue", flushQueueIndex)
 	}
 }
@@ -279,6 +278,8 @@ func (i *Ingester) flushUserSeries(flushQueueIndex int, userID string, fp model.
 		return nil
 	}
 
+	i.metrics.flushedSeries.WithLabelValues(reason.String()).Inc()
+
 	// flush the chunks without locking the series, as we don't want to hold the series lock for the duration of the dynamo/s3 rpcs.
 	ctx, cancel := context.WithTimeout(context.Background(), i.cfg.FlushOpTimeout)
 	defer cancel() // releases resources if slowOperation completes before timeout elapses
diff --git a/pkg/ingester/metrics.go b/pkg/ingester/metrics.go
index 91adcaed3b1..52ff53ad3cd 100644
--- a/pkg/ingester/metrics.go
+++ b/pkg/ingester/metrics.go
@@ -52,7 +52,7 @@ type ingesterMetrics struct {
 	chunkSize                     prometheus.Histogram
 	chunkAge                      prometheus.Histogram
 	memoryChunks                  prometheus.Gauge
-	flushReasons                  *prometheus.CounterVec
+	flushedSeries                 *prometheus.CounterVec
 	droppedChunks                 prometheus.Counter
 	oldestUnflushedChunkTimestamp prometheus.Gauge
 }
@@ -187,9 +187,9 @@ func newIngesterMetrics(r prometheus.Registerer, createMetricsConflictingWithTSD
 			Name: "cortex_ingester_memory_chunks",
 			Help: "The total number of chunks in memory.",
 		}),
-		flushReasons: promauto.With(r).NewCounterVec(prometheus.CounterOpts{
-			Name: "cortex_ingester_flush_reasons",
-			Help: "Total number of series scheduled for flushing, with reasons.",
+		flushedSeries: promauto.With(r).NewCounterVec(prometheus.CounterOpts{
+			Name: "cortex_ingester_series_flushed_total",
+			Help: "Total number of flushed series, with reasons.",
 		}, []string{"reason"}),
 		droppedChunks: promauto.With(r).NewCounter(prometheus.CounterOpts{
 			Name: "cortex_ingester_dropped_chunks_total",