diff --git a/crates/milli/src/update/new/extract/searchable/extract_word_pair_proximity_docids.rs b/crates/milli/src/update/new/extract/searchable/extract_word_pair_proximity_docids.rs
index dcd9e3a78..db64e1132 100644
--- a/crates/milli/src/update/new/extract/searchable/extract_word_pair_proximity_docids.rs
+++ b/crates/milli/src/update/new/extract/searchable/extract_word_pair_proximity_docids.rs
@@ -42,8 +42,6 @@ impl SearchableExtractor for WordPairProximityDocidsExtractor {
         let rtxn = &context.rtxn;
 
         let mut key_buffer = bumpalo::collections::Vec::new_in(doc_alloc);
-        let mut del_word_pair_proximity = bumpalo::collections::Vec::new_in(doc_alloc);
-        let mut add_word_pair_proximity = bumpalo::collections::Vec::new_in(doc_alloc);
 
         let mut new_fields_ids_map = context.new_fields_ids_map.borrow_mut_or_yield();
         let new_fields_ids_map = &mut *new_fields_ids_map;
@@ -65,7 +63,8 @@ impl SearchableExtractor for WordPairProximityDocidsExtractor {
                     new_fields_ids_map,
                     &mut word_positions,
                     &mut |(w1, w2), prox| {
-                        del_word_pair_proximity.push(((w1, w2), prox));
+                        let key = build_key(prox, &w1, &w2, &mut key_buffer);
+                        cached_sorter.insert_del_u32(key, docid)
                     },
                 )?;
             }
@@ -77,7 +76,8 @@ impl SearchableExtractor for WordPairProximityDocidsExtractor {
                     new_fields_ids_map,
                     &mut word_positions,
                     &mut |(w1, w2), prox| {
-                        del_word_pair_proximity.push(((w1, w2), prox));
+                        let key = build_key(prox, &w1, &w2, &mut key_buffer);
+                        cached_sorter.insert_del_u32(key, docid)
                     },
                 )?;
                 let document = inner.merged(rtxn, index, context.db_fields_ids_map)?;
@@ -87,7 +87,8 @@ impl SearchableExtractor for WordPairProximityDocidsExtractor {
                     new_fields_ids_map,
                     &mut word_positions,
                     &mut |(w1, w2), prox| {
-                        add_word_pair_proximity.push(((w1, w2), prox));
+                        let key = build_key(prox, &w1, &w2, &mut key_buffer);
+                        cached_sorter.insert_add_u32(key, docid)
                     },
                 )?;
             }
@@ -99,25 +100,13 @@ impl SearchableExtractor for WordPairProximityDocidsExtractor {
                     new_fields_ids_map,
                     &mut word_positions,
                     &mut |(w1, w2), prox| {
-                        add_word_pair_proximity.push(((w1, w2), prox));
+                        let key = build_key(prox, &w1, &w2, &mut key_buffer);
+                        cached_sorter.insert_add_u32(key, docid)
                     },
                 )?;
             }
         }
 
-        del_word_pair_proximity.sort_unstable();
-        del_word_pair_proximity.dedup_by(|(k1, _), (k2, _)| k1 == k2);
-        for ((w1, w2), prox) in del_word_pair_proximity.iter() {
-            let key = build_key(*prox, w1, w2, &mut key_buffer);
-            cached_sorter.insert_del_u32(key, docid)?;
-        }
-
-        add_word_pair_proximity.sort_unstable();
-        add_word_pair_proximity.dedup_by(|(k1, _), (k2, _)| k1 == k2);
-        for ((w1, w2), prox) in add_word_pair_proximity.iter() {
-            let key = build_key(*prox, w1, w2, &mut key_buffer);
-            cached_sorter.insert_add_u32(key, docid)?;
-        }
         Ok(())
     }
 }
@@ -138,24 +127,26 @@ fn build_key<'a>(
 
 fn word_positions_into_word_pair_proximity(
     word_positions: &mut VecDeque<(Rc<str>, u16)>,
-    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8),
-) {
+    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8) -> Result<()>,
+) -> Result<()> {
     let (head_word, head_position) = word_positions.pop_front().unwrap();
     for (word, position) in word_positions.iter() {
         let prox = index_proximity(head_position as u32, *position as u32) as u8;
         if prox > 0 && prox < MAX_DISTANCE as u8 {
-            word_pair_proximity((head_word.clone(), word.clone()), prox);
+            word_pair_proximity((head_word.clone(), word.clone()), prox)?;
         }
     }
+    Ok(())
 }
 
 fn drain_word_positions(
     word_positions: &mut VecDeque<(Rc<str>, u16)>,
-    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8),
-) {
+    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8) -> Result<()>,
+) -> Result<()> {
     while !word_positions.is_empty() {
-        word_positions_into_word_pair_proximity(word_positions, word_pair_proximity);
+        word_positions_into_word_pair_proximity(word_positions, word_pair_proximity)?;
     }
+    Ok(())
 }
 
 fn process_document_tokens<'doc>(
@@ -163,20 +154,20 @@ fn process_document_tokens<'doc>(
     document_tokenizer: &DocumentTokenizer,
     fields_ids_map: &mut GlobalFieldsIdsMap,
     word_positions: &mut VecDeque<(Rc<str>, u16)>,
-    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8),
+    word_pair_proximity: &mut impl FnMut((Rc<str>, Rc<str>), u8) -> Result<()>,
 ) -> Result<()> {
     let mut field_id = None;
     let mut token_fn = |_fname: &str, fid: FieldId, pos: u16, word: &str| {
         if field_id != Some(fid) {
             field_id = Some(fid);
-            drain_word_positions(word_positions, word_pair_proximity);
+            drain_word_positions(word_positions, word_pair_proximity)?;
         }
         // drain the proximity window until the head word is considered close to the word we are inserting.
         while word_positions
             .front()
             .map_or(false, |(_w, p)| index_proximity(*p as u32, pos as u32) >= MAX_DISTANCE)
         {
-            word_positions_into_word_pair_proximity(word_positions, word_pair_proximity);
+            word_positions_into_word_pair_proximity(word_positions, word_pair_proximity)?;
         }
 
         // insert the new word.
@@ -185,6 +176,6 @@ fn process_document_tokens<'doc>(
     };
     document_tokenizer.tokenize_document(document, fields_ids_map, &mut token_fn)?;
 
-    drain_word_positions(word_positions, word_pair_proximity);
+    drain_word_positions(word_positions, word_pair_proximity)?;
     Ok(())
 }