diff --git a/lib/src/main/java/graphql/nadel/result/NadelResultTracker.kt b/lib/src/main/java/graphql/nadel/result/NadelResultTracker.kt
index 8aaee080b..905a8dbbb 100644
--- a/lib/src/main/java/graphql/nadel/result/NadelResultTracker.kt
+++ b/lib/src/main/java/graphql/nadel/result/NadelResultTracker.kt
@@ -20,6 +20,21 @@ internal class NadelResultTracker {
         DeadEnd,
     }
 
+    /**
+     * So… in Nadel the result can change a lot.
+     *
+     * This function lets you track where a result node went to in the overall response sent to the user.
+     *
+     * I haven't benchmarked this yet, but in theory it should be more performant than say using
+     * [graphql.nadel.engine.transform.result.json.JsonNodes].
+     *
+     * In the past we used to track the [NadelResultPathSegment]s for each [JsonNode] but that was horrible
+     * performance wise because we created one List for each result node
+     * i.e. as the result grew, both in depth and result node count, you'd allocate tons of (big) lists.
+     *
+     * This implementation keeps track of the _current_ [NadelResultPathSegment]s and returns that if it
+     * finds the [node] in question.
+     */
     suspend fun getResultPath(
         queryPath: NadelQueryPath,
         node: JsonNode,