diff --git a/examples/xgboost-comprehensive/README.md b/examples/xgboost-comprehensive/README.md
index d28b719edbf2..15dcb03b2811 100644
--- a/examples/xgboost-comprehensive/README.md
+++ b/examples/xgboost-comprehensive/README.md
@@ -18,7 +18,7 @@ It differs from the [xgboost-quickstart](https://github.com/adap/flower/tree/mai
 
 ## Training Strategies
 
-This example provides two training strategies, **bagging aggregation** and **cyclic training**.
+This example provides two training strategies, [**bagging aggregation**](https://flower.ai/docs/framework/tutorial-quickstart-xgboost.html#tree-based-bagging-aggregation) and [**cyclic training**](https://flower.ai/docs/framework/tutorial-quickstart-xgboost.html#cyclic_training).
 
 ### Bagging Aggregation
 
diff --git a/examples/xgboost-comprehensive/xgboost_comprehensive/server_app.py b/examples/xgboost-comprehensive/xgboost_comprehensive/server_app.py
index 07bbd3cb3ea4..a616d2a15165 100644
--- a/examples/xgboost-comprehensive/xgboost_comprehensive/server_app.py
+++ b/examples/xgboost-comprehensive/xgboost_comprehensive/server_app.py
@@ -77,6 +77,11 @@ def evaluate_fn(
             )
             auc = round(float(eval_results.split("\t")[1].split(":")[1]), 4)
 
+            # Save results to disk.
+            # Note we add new entry to the same file with each call to this function.
+            with open(f"./centralised_eval.txt", "a", encoding="utf-8") as fp:
+                fp.write(f"Round:{server_round},AUC:{auc}\n")
+
             return 0, {"AUC": auc}
 
     return evaluate_fn