diff --git a/presto-native-execution/presto_cpp/main/operators/ShuffleRead.cpp b/presto-native-execution/presto_cpp/main/operators/ShuffleRead.cpp
index 9d049ec037442..bc85443dea742 100644
--- a/presto-native-execution/presto_cpp/main/operators/ShuffleRead.cpp
+++ b/presto-native-execution/presto_cpp/main/operators/ShuffleRead.cpp
@@ -14,7 +14,6 @@
 #include "presto_cpp/main/operators/ShuffleRead.h"
 #include "presto_cpp/main/operators/ShuffleExchangeSource.h"
 #include "velox/common/Casts.h"
-#include "velox/exec/Exchange.h"
 #include "velox/row/CompactRow.h"
 
 using namespace facebook::velox::exec;
@@ -25,61 +24,30 @@ velox::core::PlanNodeId deserializePlanNodeId(const folly::dynamic& obj) {
   return obj["id"].asString();
 }
 
-namespace {
-class ShuffleRead : public Exchange {
- public:
-  ShuffleRead(
-      int32_t operatorId,
-      DriverCtx* ctx,
-      const std::shared_ptr<const ShuffleReadNode>& shuffleReadNode,
-      std::shared_ptr<ExchangeClient> exchangeClient)
-      : Exchange(
-            operatorId,
-            ctx,
-            std::make_shared<core::ExchangeNode>(
-                shuffleReadNode->id(),
-                shuffleReadNode->outputType(),
-                VectorSerde::Kind::kCompactRow),
-            exchangeClient,
-            "ShuffleRead") {
-    initStats();
-  }
-
-  RowVectorPtr getOutput() override;
-
-  void close() override;
-
- protected:
-  VectorSerde* getSerde() override {
-    VELOX_UNSUPPORTED("ShuffleReadOperator doesn't use serde");
-  }
-
- private:
-  static inline const std::string kShuffleDecodeTime{"shuffleDecodeWallNanos"};
-  static inline const std::string kShuffleNumBatchesPerRead{
-      "shuffleNumBatchesPerRead"};
-  static inline const std::string kShuffleNumBatches{"shuffleNumBatches"};
-
-  void initStats();
-
-  void resetOutputState();
-
-  int64_t numBatches_{0};
-  std::unordered_map<std::string, velox::RuntimeMetric> runtimeStats_;
-
-  size_t nextRow_{0};
-  size_t nextPage_{0};
-  // Reusable buffers.
-  std::vector<std::string_view> rows_;
-  std::vector<size_t> pageRows_;
-};
+ShuffleRead::ShuffleRead(
+    int32_t operatorId,
+    DriverCtx* ctx,
+    const std::shared_ptr<const ShuffleReadNode>& shuffleReadNode,
+    std::shared_ptr<ExchangeClient> exchangeClient)
+    : Exchange(
+          operatorId,
+          ctx,
+          std::make_shared<core::ExchangeNode>(
+              shuffleReadNode->id(),
+              shuffleReadNode->outputType(),
+              VectorSerde::Kind::kCompactRow),
+          exchangeClient,
+          "ShuffleRead") {
+  initStats();
+}
 
 void ShuffleRead::initStats() {
   VELOX_CHECK(runtimeStats_.empty());
   runtimeStats_.insert(
       std::pair{kShuffleDecodeTime, velox::RuntimeCounter::Unit::kNanos});
   runtimeStats_.insert(
-      std::pair{kShuffleNumBatchesPerRead, velox::RuntimeCounter::Unit::kNone});
+      std::pair{
+          kShufflePagesPerInputBatch, velox::RuntimeCounter::Unit::kNone});
 }
 
 void ShuffleRead::resetOutputState() {
@@ -122,8 +90,8 @@ RowVectorPtr ShuffleRead::getOutput() {
       }
     }
     if (!currentPages_.empty()) {
-      runtimeStats_[kShuffleNumBatchesPerRead].addValue(currentPages_.size());
-      numBatches_ += currentPages_.size();
+      runtimeStats_[kShufflePagesPerInputBatch].addValue(currentPages_.size());
+      ++numInputBatches_;
     }
   }
   VELOX_CHECK_LE(nextRow_, rows_.size());
@@ -177,12 +145,11 @@ void ShuffleRead::close() {
     }
     lockedStats->runtimeStats[name] = metric;
   }
-  if (numBatches_ != 0) {
+  if (numInputBatches_ != 0) {
     lockedStats->addRuntimeStat(
-        kShuffleNumBatches, RuntimeCounter(numBatches_));
+        kShuffleInputBatches, RuntimeCounter(numInputBatches_));
   }
 }
-} // namespace
 
 folly::dynamic ShuffleReadNode::serialize() const {
   auto obj = PlanNode::serialize();
diff --git a/presto-native-execution/presto_cpp/main/operators/ShuffleRead.h b/presto-native-execution/presto_cpp/main/operators/ShuffleRead.h
index 33a410151eb60..fbda62cf2e0e1 100644
--- a/presto-native-execution/presto_cpp/main/operators/ShuffleRead.h
+++ b/presto-native-execution/presto_cpp/main/operators/ShuffleRead.h
@@ -14,6 +14,7 @@
 #pragma once
 
 #include "velox/core/PlanNode.h"
+#include "velox/exec/Exchange.h"
 #include "velox/exec/Operator.h"
 
 namespace facebook::presto::operators {
@@ -90,6 +91,43 @@ class ShuffleReadNode : public velox::core::PlanNode {
   const velox::RowTypePtr outputType_;
 };
 
+class ShuffleRead : public velox::exec::Exchange {
+ public:
+  ShuffleRead(
+      int32_t operatorId,
+      velox::exec::DriverCtx* ctx,
+      const std::shared_ptr<const ShuffleReadNode>& shuffleReadNode,
+      std::shared_ptr<velox::exec::ExchangeClient> exchangeClient);
+
+  velox::RowVectorPtr getOutput() override;
+
+  void close() override;
+
+  static inline const std::string kShuffleDecodeTime{"shuffleDecodeWallNanos"};
+  static inline const std::string kShufflePagesPerInputBatch{
+      "shuffleNumPagesPerInputBatch"};
+  static inline const std::string kShuffleInputBatches{"shuffleInputBatches"};
+
+ protected:
+  velox::VectorSerde* getSerde() override {
+    VELOX_UNSUPPORTED("ShuffleReadOperator doesn't use serde");
+  }
+
+ private:
+  void initStats();
+
+  void resetOutputState();
+
+  int64_t numInputBatches_{0};
+  std::unordered_map<std::string, velox::RuntimeMetric> runtimeStats_;
+
+  size_t nextRow_{0};
+  size_t nextPage_{0};
+  // Reusable buffers.
+  std::vector<std::string_view> rows_;
+  std::vector<size_t> pageRows_;
+};
+
 class ShuffleReadTranslator : public velox::exec::Operator::PlanNodeTranslator {
  public:
   std::unique_ptr<velox::exec::Operator> toOperator(
diff --git a/presto-native-execution/presto_cpp/main/operators/tests/ShuffleTest.cpp b/presto-native-execution/presto_cpp/main/operators/tests/ShuffleTest.cpp
index b04b940ec6c3e..46b6bb70230bb 100644
--- a/presto-native-execution/presto_cpp/main/operators/tests/ShuffleTest.cpp
+++ b/presto-native-execution/presto_cpp/main/operators/tests/ShuffleTest.cpp
@@ -1470,21 +1470,23 @@ TEST_F(ShuffleTest, shuffleReadRuntimeStats) {
         taskCursor->task()->taskStats().pipelineStats[0].operatorStats[0];
     const auto& runtimeStats = operatorStats.runtimeStats;
 
-    ASSERT_EQ(runtimeStats.count("shuffleDecodeWallNanos"), 1);
-    const auto& decodeTimeStat = runtimeStats.at("shuffleDecodeWallNanos");
+    ASSERT_EQ(runtimeStats.count(ShuffleRead::kShuffleDecodeTime), 1);
+    const auto& decodeTimeStat =
+        runtimeStats.at(ShuffleRead::kShuffleDecodeTime);
     ASSERT_GT(decodeTimeStat.count, 0);
     ASSERT_GT(decodeTimeStat.sum, 0);
     ASSERT_EQ(velox::RuntimeCounter::Unit::kNanos, decodeTimeStat.unit);
 
-    ASSERT_EQ(runtimeStats.count("shuffleNumBatchesPerRead"), 1);
+    ASSERT_EQ(runtimeStats.count(ShuffleRead::kShufflePagesPerInputBatch), 1);
     const auto& batchesPerReadStat =
-        runtimeStats.at("shuffleNumBatchesPerRead");
+        runtimeStats.at(ShuffleRead::kShufflePagesPerInputBatch);
     ASSERT_EQ(velox::RuntimeCounter::Unit::kNone, batchesPerReadStat.unit);
     ASSERT_GT(batchesPerReadStat.count, 0);
     ASSERT_GT(batchesPerReadStat.sum, 0);
 
-    ASSERT_EQ(runtimeStats.count("shuffleNumBatches"), 1);
-    const auto& numBatchesStat = runtimeStats.at("shuffleNumBatches");
+    ASSERT_EQ(runtimeStats.count(ShuffleRead::kShuffleInputBatches), 1);
+    const auto& numBatchesStat =
+        runtimeStats.at(ShuffleRead::kShuffleInputBatches);
     ASSERT_GT(numBatchesStat.count, 0);
     ASSERT_GT(numBatchesStat.sum, 0);
     ASSERT_EQ(velox::RuntimeCounter::Unit::kNone, numBatchesStat.unit);