diff --git a/src/contracts/mod.rs b/src/contracts/mod.rs
index 84737307..c451eade 100644
--- a/src/contracts/mod.rs
+++ b/src/contracts/mod.rs
@@ -146,10 +146,14 @@ impl IdentityManager {
         self.tree_depth
     }
 
-    pub async fn max_batch_size(&self) -> usize {
+    pub async fn max_insertion_batch_size(&self) -> usize {
         self.insertion_prover_map.read().await.max_batch_size()
     }
 
+    pub async fn max_deletion_batch_size(&self) -> usize {
+        self.deletion_prover_map.read().await.max_batch_size()
+    }
+
     #[must_use]
     pub const fn initial_leaf_value(&self) -> Field {
         self.initial_leaf_value
diff --git a/src/task_monitor/tasks/delete_identities.rs b/src/task_monitor/tasks/delete_identities.rs
index 83b8a984..53905eeb 100644
--- a/src/task_monitor/tasks/delete_identities.rs
+++ b/src/task_monitor/tasks/delete_identities.rs
@@ -9,6 +9,7 @@ use tracing::info;
 use crate::database::types::DeletionEntry;
 use crate::database::Database;
 use crate::identity_tree::{Hash, Latest, TreeVersion};
+use crate::prover::{ProverConfiguration, ProverType};
 
 pub struct DeleteIdentities {
     database:                Arc<Database>,
@@ -61,9 +62,6 @@ async fn delete_identities(
     loop {
         let deletions = database.get_deletions().await?;
         if deletions.is_empty() {
-            // Sleep for one hour
-            // TODO: should we make this dynamic? This causes an issue with tests so its set
-            // to 1 sec atm
             tokio::time::sleep(tokio::time::Duration::from_secs(5)).await;
             continue;
         }
@@ -78,6 +76,22 @@ async fn delete_identities(
             // Dedup deletion entries
             let deletions = deletions.into_iter().collect::<HashSet<DeletionEntry>>();
 
+            // Get the max batch size for deletion provers
+            let max_batch_size = database
+                .get_provers()
+                .await?
+                .iter()
+                .filter(|p| p.prover_type == ProverType::Deletion)
+                .max_by_key(|p| p.batch_size)
+                .map(|p| p.batch_size)
+                .expect("Could not get match batch size");
+
+            // Trim deletions the max deletion batch size available
+            let deletions = deletions
+                .into_iter()
+                .take(max_batch_size)
+                .collect::<HashSet<DeletionEntry>>();
+
             let (leaf_indices, previous_commitments): (Vec<usize>, Vec<Hash>) = deletions
                 .iter()
                 .map(|d| (d.leaf_index, d.commitment))
diff --git a/src/task_monitor/tasks/process_identities.rs b/src/task_monitor/tasks/process_identities.rs
index f9e57460..85834477 100644
--- a/src/task_monitor/tasks/process_identities.rs
+++ b/src/task_monitor/tasks/process_identities.rs
@@ -72,7 +72,9 @@ async fn process_identities(
     identity_manager.await_clean_slate().await?;
 
     info!("Starting identity processor.");
-    let batch_size = identity_manager.max_batch_size().await;
+    let insertion_batch_size = identity_manager.max_insertion_batch_size().await;
+    let deletion_batch_size = identity_manager.max_deletion_batch_size().await;
+    let batch_size = std::cmp::max(insertion_batch_size, deletion_batch_size);
 
     // We start a timer and force it to perform one initial tick to avoid an
     // immediate trigger.
@@ -101,6 +103,9 @@ async fn process_identities(
                 // If the timer has fired we want to insert whatever
                 // identities we have, even if it's not many. This ensures
                 // a minimum quality of service for API users.
+
+                batching_tree.
+
                 let updates = batching_tree.peek_next_updates(batch_size);
                 if updates.is_empty() {
                     continue;