diff --git a/etl/eligibility/ha_15_32/ha_analysis_batch_3.py b/etl/eligibility/ha_15_32/ha_analysis_batch_3.py
index 1c320f9c..3341e34c 100644
--- a/etl/eligibility/ha_15_32/ha_analysis_batch_3.py
+++ b/etl/eligibility/ha_15_32/ha_analysis_batch_3.py
@@ -2965,6 +2965,14 @@ def forecast_remaining_sales(loader):
         gbis_remaining = int(np.round(gbis_remaining * ha_gbis_sale_conversion))
         gbis_remaining_revenue = int(gbis_remaining * gbis_rate)
 
+        # GBIS delta
+        if original_warmfront_gbis == 0:
+            gbis_delta_vs_original_estimate = 100 * gbis_total
+        else:
+            gbis_delta_vs_original_estimate = 100 * (
+                gbis_total - original_warmfront_gbis
+            ) / original_warmfront_gbis
+
         to_append = {
             ("", "", "", "HA Name"): ha_name,
             # ECO4 - original warmfront figures
@@ -2987,7 +2995,7 @@ def forecast_remaining_sales(loader):
                 "ECO4 - post CIGA - #"],
             ("ECO4 post-ciga", "", "Estimated total eligible - £", ""): eco4_post_ciga_total_results[
                 "ECO4 - post CIGA - £"],
-            ("ECO4 post-ciga", "", "Delta vs original estimate", ""): eco4_delta_vs_original_estimate,
+            ("ECO4 post-ciga", "", "Delta vs original estimate - %", ""): eco4_delta_vs_original_estimate,
             # ECO4 - asset list, post ciga, remaining
             ("ECO4 post-ciga", "", "Estimated remaining eligible - #", ""): eco4_post_ciga_remaining_results[
                 "ECO4 - post CIGA - #"],
@@ -3021,14 +3029,15 @@ def forecast_remaining_sales(loader):
                 "Estimated CIGA failures - £"
             ],
             # GBIS postcode list
-            ("", "Warmfront post code list", "Total - #", "GBIS total"): gbis_total,
-            ("", "Warmfront post code list", "Remaining - #", "GBIS total"): gbis_remaining,
-            ("", "Warmfront post code list", "Total - £", "GBIS total"): gbis_total_revenue,
-            ("", "Warmfront post code list", "Remaining - £", "GBIS total"): gbis_remaining_revenue,
+            ("GBIS Postcode list", "Warmfront post code list", "Total - #", "GBIS total"): gbis_total,
+            ("GBIS Postcode list", "Warmfront post code list", "Total - £", "GBIS total"): gbis_total_revenue,
+            ("GBIS Postcode list", "", "Delta vs original estimate - %", ""): gbis_delta_vs_original_estimate,
+            ("GBIS Postcode list", "Warmfront post code list", "Remaining - #", "GBIS total"): gbis_remaining,
+            ("GBIS Postcode list", "Warmfront post code list", "Remaining - £", "GBIS total"): gbis_remaining_revenue,
         }
 
         # Make sure nothing is forgotten due to duplicate multi-index keys
-        if len(to_append) != 32:
+        if len(to_append) != 33:
             raise ValueError("Something went wrong")
 
         results.append(to_append)
@@ -3039,11 +3048,31 @@ def forecast_remaining_sales(loader):
     for col in results.columns:
         if col == ('', '', '', 'HA Name'):
             totals_row[col] = "Total"
-        elif col == ("ECO4 post-ciga", "", "Delta vs original estimate", ""):
-            totals_row[col] = results[col].mean()
+        elif col in [
+            ("ECO4 post-ciga", "", "Delta vs original estimate - %", ""),
+            ("GBIS Postcode list", "", "Delta vs original estimate - %", "")
+        ]:
+            totals_row[col] = None
         else:
             totals_row[col] = results[col].sum()
 
+    # For the delta columns, we calculate the delta on the totals
+    totals_row[("ECO4 post-ciga", "", "Delta vs original estimate - %", "")] = round(
+        100 * (
+            totals_row[("ECO4 post-ciga", "", "Estimated total eligible - #", "")] -
+            totals_row[("", "Original Warmfront estimate", "Total - #", "ECO4 - November")]
+        ) / totals_row[("", "Original Warmfront estimate", "Total - #", "ECO4 - November")],
+        1
+    )
+
+    totals_row[("GBIS Postcode list", "", "Delta vs original estimate - %", "")] = round(
+        100 * (
+            totals_row[("GBIS Postcode list", "Warmfront post code list", "Total - #", "GBIS total")] -
+            totals_row[("", "Original Warmfront estimate", "Total - #", "GBIS - November")]
+        ) / totals_row[("", "Original Warmfront estimate", "Total - #", "GBIS - November")],
+        1
+    )
+
     blank_row = pd.DataFrame([{col: "" for col in results.columns}])
 
     # Put together a Warmfront original remaining ECO4 vs asset list remaining ECO4 and same for GBIS, as well as totals
@@ -3204,10 +3233,35 @@ def forecast_remaining_sales(loader):
     ]
 
     results = pd.concat(
-        [results, pd.DataFrame([headlines]), pd.DataFrame([totals_row]), blank_row, blank_row,
-         pd.DataFrame(assumptions)]
+        [
+            results,
+            pd.DataFrame([totals_row]),
+            pd.DataFrame(headlines),
+            blank_row,
+            blank_row,
+            pd.DataFrame(assumptions)
+        ]
     )
 
+    # header_rows = [
+    #     [name[0] for name in results.columns.values],
+    #     [name[1] for name in results.columns.values],
+    #     [name[2] for name in results.columns.values],
+    #     [name[3] for name in results.columns.values]
+    # ]
+
+    # Step 2: Write the transformed header and DataFrame data to CSV.
+    # Open the file in write mode.
+    import csv
+    with open("HA Remaining Analysis.csv", "w", newline="") as file:
+        # writer = csv.writer(file)
+
+        # Write the header rows.
+        # writer.writerows(header_rows)
+
+        # Write the DataFrame data without the index (adjust if you want the index).
+        results.to_csv(file, header=True, index=False)
+
 
 def app():
     """