diff --git a/pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py b/pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py
index fa6031f..6e07728 100644
--- a/pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py
+++ b/pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py
@@ -83,8 +83,13 @@ class EESDatasetStream(Stream):
         with zf.open(target) as f:
             df = pd.read_csv(f, dtype=str, keep_default_na=False, encoding=self._encoding)
 
-        # Strip UTF-8 BOM from column names (some DfE files have a BOM on the first column)
-        df.columns = df.columns.str.lstrip("\ufeff")
+        # Strip BOM from first column name — handles both:
+        # - UTF-8 BOM decoded as Unicode (\ufeff) when read with utf-8/utf-8-sig
+        # - UTF-8 BOM bytes decoded as Latin-1 () when read with latin-1
+        cols = list(df.columns)
+        if cols:
+            cols[0] = cols[0].lstrip("\ufeff").lstrip("")
+            df.columns = cols
 
         # Filter to school-level data if the column exists
         if "geographic_level" in df.columns:
@@ -292,7 +297,7 @@ class EESAdmissionsStream(EESDatasetStream):
     primary_keys = ["school_urn", "time_period"]
     _publication_slug = "primary-and-secondary-school-applications-and-offers"
     _target_filename = "SchoolLevel"
-    _encoding = "latin-1"
+    _encoding = "utf-8-sig"  # UTF-8 with BOM — sig variant strips the BOM automatically
     schema = th.PropertiesList(
         th.Property("time_period", th.StringType, required=True),
         th.Property("school_urn", th.StringType, required=True),