City-of-Helsinki · karisal-anders · Jan 28, 2026 · Jan 27, 2026 · Jan 27, 2026 · Jan 27, 2026
@@ -576,26 +576,33 @@ def get_queryset(self, request):
 class EmployerSummerVoucherAdmin(admin.ModelAdmin):
     list_display = [
         "id",
-        "summer_voucher_serial_number",
+        "youth_summer_voucher_id",
+        "_obsolete_unclean_serial_number",
+        "has_migrated_obsolete_serial_number",
         "target_group",
         "application__company__name",
         "created_at",
         "modified_at",
     ]
     list_filter = [
         "target_group",
+        ("youth_summer_voucher_id", admin.EmptyFieldListFilter),
+        ("_obsolete_unclean_serial_number", admin.EmptyFieldListFilter),
         "created_at",
         "modified_at",
     ]
     date_hierarchy = "created_at"
     search_fields = [
-        "summer_voucher_serial_number",
+        "youth_summer_voucher__summer_voucher_serial_number",
+        "_obsolete_unclean_serial_number",
         "id",
         "application__company__name",
     ]
-    autocomplete_fields = ["application"]
+    autocomplete_fields = [
+        "application",
+        "youth_summer_voucher",
+    ]
     readonly_fields = [
-        "summer_voucher_serial_number",
         "target_group",
         "application",
         "masked_employee_ssn",
@@ -656,8 +663,24 @@ def queryset(self, request):
             .queryset(request)
             .select_related("application")
             .select_related("application__company")
+            .select_related("youth_summer_voucher")
+        )
+
+    def has_migrated_obsolete_serial_number(self, obj):
+        old_serial = obj._obsolete_unclean_serial_number.strip()
+        # Either the old serial is empty, or it matches the current
+        # youth summer voucher ID
+        return not old_serial or (
+            old_serial.isdigit()
+            and obj.youth_summer_voucher
+            and int(old_serial) == obj.youth_summer_voucher.summer_voucher_serial_number
         )
 
+    has_migrated_obsolete_serial_number.boolean = True  # i.e. True/False icon
+    has_migrated_obsolete_serial_number.short_description = _(
+        "Has obsolete serial number been migrated?"
+    )
+
     def masked_employee_ssn(self, obj):
         """Mask employee social security number for display."""
         return mask_social_security_number(obj.employee_ssn)

@@ -209,6 +209,13 @@ class EmployerSummerVoucherSerializer(serializers.ModelSerializer):
         many=True,
         help_text="Attachments of the application (read-only)",
     )
+    # Backward compatibility field for frontend using
+    # EmployerSummerVoucher summer_voucher_serial_number property and its setter:
+    summer_voucher_serial_number = serializers.CharField(
+        max_length=256,
+        allow_blank=True,
+        required=False,
+    )
 
     class Meta:
         model = EmployerSummerVoucher

@@ -39,6 +39,7 @@ class ExcelField(NamedTuple):
 INVOICER_EMAIL_FIELD_TITLE = _("Laskuttajan sähköposti")
 INVOICER_NAME_FIELD_TITLE = _("Laskuttajan nimi")
 INVOICER_PHONE_NUMBER_FIELD_TITLE = _("Laskuttajan Puhelin")
+VOUCHER_NUMBER_FIELD_TITLE = _("Setelin numero")
 
 REMOVABLE_REPORTING_FIELD_TITLES = [
     _("Y-tunnus"),
@@ -105,7 +106,7 @@ class ExcelField(NamedTuple):
         APPLICATION_LANGUAGE_FIELD_TITLE, "%s", ["application__language"], 15, "white"
     ),
     ExcelField(
-        _("Setelin numero"), "%s", ["summer_voucher_serial_number"], 30, "white"
+        VOUCHER_NUMBER_FIELD_TITLE, "%s", ["summer_voucher_serial_number"], 30, "white"
     ),
     ExcelField(
         SPECIAL_CASE_FIELD_TITLE,

@@ -0,0 +1,70 @@
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("applications", "0040_alter_employerapplication_options_and_more"),
+    ]
+
+    operations = [
+        migrations.RenameField(
+            model_name="employersummervoucher",
+            old_name="summer_voucher_serial_number",
+            new_name="_obsolete_unclean_serial_number",
+        ),
+        migrations.RenameField(
+            model_name="historicalemployersummervoucher",
+            old_name="summer_voucher_serial_number",
+            new_name="_obsolete_unclean_serial_number",
+        ),
+        migrations.AlterField(
+            model_name="employersummervoucher",
+            name="_obsolete_unclean_serial_number",
+            field=models.CharField(
+                blank=True,
+                help_text="Old obsolete unclean summer_voucher_serial_number values before data migration in early 2026. Can be used for manual data cleaning and as fallback summer_voucher_serial_number values in historical data.",
+                max_length=256,
+                verbose_name="obsolete unclean summer voucher serial number",
+            ),
+        ),
+        migrations.AlterField(
+            model_name="historicalemployersummervoucher",
+            name="_obsolete_unclean_serial_number",
+            field=models.CharField(
+                blank=True,
+                help_text="Old obsolete unclean summer_voucher_serial_number values before data migration in early 2026. Can be used for manual data cleaning and as fallback summer_voucher_serial_number values in historical data.",
+                max_length=256,
+                verbose_name="obsolete unclean summer voucher serial number",
+            ),
+        ),
+        migrations.AddField(
+            model_name="employersummervoucher",
+            name="youth_summer_voucher",
+            field=models.ForeignKey(
+                db_column="summer_voucher_serial_number",
+                null=True,
+                blank=True,
+                on_delete=django.db.models.deletion.PROTECT,
+                related_name="employer_summer_vouchers",
+                to="applications.youthsummervoucher",
+                to_field="summer_voucher_serial_number",
+                verbose_name="youth summer voucher",
+            ),
+        ),
+        migrations.AddField(
+            model_name="historicalemployersummervoucher",
+            name="youth_summer_voucher",
+            field=models.ForeignKey(
+                blank=True,
+                db_column="summer_voucher_serial_number",
+                db_constraint=False,
+                null=True,
+                on_delete=django.db.models.deletion.DO_NOTHING,
+                related_name="+",
+                to="applications.youthsummervoucher",
+                to_field="summer_voucher_serial_number",
+                verbose_name="youth summer voucher",
+            ),
+        ),
+    ]
@@ -0,0 +1,43 @@
+from django.db import migrations
+
+from applications.migrations.helpers.serial_number_foreign_keys import (
+    set_current_valid_serial_number_based_foreign_keys,
+    set_historical_serial_number_based_foreign_keys,
+)
+
+
+def set_current_and_historical_serial_number_based_foreign_keys(apps, schema_editor):
+    """
+    Set youth_summer_voucher_id ForeignKey values in EmployerSummerVoucher and
+    HistoricalEmployerSummerVoucher models based on the _obsolete_unclean_serial_number
+    values.
+    """
+    employer_summer_voucher_model = apps.get_model(
+        "applications", "EmployerSummerVoucher"
+    )
+    youth_summer_voucher_model = apps.get_model("applications", "YouthSummerVoucher")
+    historical_employer_summer_voucher_model = apps.get_model(
+        "applications", "HistoricalEmployerSummerVoucher"
+    )
+    set_current_valid_serial_number_based_foreign_keys(
+        employer_summer_voucher_model, youth_summer_voucher_model
+    )
+    set_historical_serial_number_based_foreign_keys(
+        historical_employer_summer_voucher_model
+    )
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        (
+            "applications",
+            "0041_rename_serial_number_field_and_add_youth_voucher_foreign_key",
+        ),
+    ]
+
+    operations = [
+        migrations.RunPython(
+            set_current_and_historical_serial_number_based_foreign_keys,
+            migrations.RunPython.noop,
+        ),
+    ]
@@ -0,0 +1,122 @@
+import logging
+from collections import defaultdict
+
+from django.db.models import PositiveBigIntegerField
+from django.db.models.functions import Cast, Trim
+from stdnum.fi.hetu import is_valid as is_valid_finnish_social_security_number
+
+LOGGER = logging.getLogger(__name__)
+
+
+def set_current_valid_serial_number_based_foreign_keys(
+    employer_summer_voucher_model, youth_summer_voucher_model
+):
+    """
+    Convert valid serial number strings in _obsolete_unclean_serial_number to actual valid
+    ForeignKey references in youth_summer_voucher_id in EmployerSummerVoucher model.
+
+    Matching is done first by purely numeric serial numbers, and if that fails, by matching
+    social security number and creation year.
+    """
+    # Real data sizes in production on 2026-01-27 for memory and performance context:
+    # ~8k EmployerSummerVoucher objects
+    # ~17k YouthSummerVoucher objects
+    # ~19k YouthApplication objects
+    total_count = matched_by_ssn_count = matched_by_serial_count = 0
+
+    # Load all YouthSummerVoucher objects into memory for efficient lookup:
+    youth_vouchers = list(
+        youth_summer_voucher_model.objects.select_related("youth_application").only(
+            "id",
+            "summer_voucher_serial_number",
+            "youth_application__social_security_number",
+            "youth_application__encrypted_social_security_number",
+            "youth_application__created_at",
+        )
+    )
+
+    # Create a mapping from unique serial numbers to YouthSummerVoucher for quick lookup:
+    serial_to_youth_voucher = {
+        v.summer_voucher_serial_number: v
+        for v in youth_vouchers
+        if v.summer_voucher_serial_number
+    }
+
+    # Create a mapping from valid social security numbers to YouthSummerVouchers for quick lookup:
+    ssn_to_youth_vouchers = defaultdict(list)
+    for v in youth_vouchers:
+        ssn = v.youth_application.social_security_number
+        if is_valid_finnish_social_security_number(ssn):
+            ssn_to_youth_vouchers[ssn].append(v)
+
+    employer_vouchers_to_update = []
+
+    # Try to find matching YouthSummerVoucher for each EmployerSummerVoucher
+    for employer_voucher in employer_summer_voucher_model.objects.all().iterator(
+        chunk_size=1000
+    ):
+        total_count += 1
+        serial_number = employer_voucher._obsolete_unclean_serial_number.strip()
+
+        # Purely numeric serial numbers? These should be the majority
+        if serial_number.isdigit() and (
+            youth_voucher := serial_to_youth_voucher.get(int(serial_number))
+        ):
+            employer_voucher.youth_summer_voucher = youth_voucher
+            employer_vouchers_to_update.append(employer_voucher)
+            matched_by_serial_count += 1
+        else:  # Try matching by social security number & application year
+            youth_vouchers = [
+                v
+                for v in ssn_to_youth_vouchers.get(employer_voucher.employee_ssn, [])
+                if v.youth_application.created_at.year
+                == employer_voucher.created_at.year
+                and v.youth_application.created_at <= employer_voucher.created_at
+            ]
+            # There should be only at most one youth voucher per SSN per year
+            if len(youth_vouchers) == 1:
+                youth_voucher = youth_vouchers[0]
+                employer_voucher.youth_summer_voucher = youth_voucher
+                employer_vouchers_to_update.append(employer_voucher)
+                matched_by_ssn_count += 1
+
+    # Bulk update all matched EmployerSummerVoucher objects
+    updated_count = employer_summer_voucher_model.objects.bulk_update(
+        employer_vouchers_to_update,
+        ["youth_summer_voucher"],
+        batch_size=500,  # To limit a single batch's SQL UPDATE clause size
+    )
+
+    # Log summary of results
+    LOGGER.info(
+        f"Handled {total_count} employer summer vouchers, updated {updated_count}:"
+    )
+    LOGGER.info(f"- Matched by voucher serial number: {matched_by_serial_count}")
+    LOGGER.info(f"- Matched by social security number & year: {matched_by_ssn_count}")
+    LOGGER.info(f"- Failed to match: {total_count - updated_count} and left as is")
+
+
+def set_historical_serial_number_based_foreign_keys(
+    historical_employer_summer_voucher_model,
+):
+    """
+    Convert non-negative integer string values in _obsolete_unclean_serial_number to actual integer
+    values in youth_summer_voucher_id in HistoricalEmployerSummerVoucher model.
+
+    Not trying to make the historical records perfect, just doing the minimum obvious conversion.
+    """
+    total_count = historical_employer_summer_voucher_model.objects.count()
+
+    # Real data in production was Jan 2026 around ~55k HistoricalEmployerSummerVoucher rows
+    updated_count = historical_employer_summer_voucher_model.objects.filter(
+        # Non-negative integer strings with possible leading/trailing whitespace:
+        _obsolete_unclean_serial_number__regex=r"^\s*[0-9]+\s*$"
+    ).update(
+        youth_summer_voucher_id=Cast(
+            Trim("_obsolete_unclean_serial_number"), PositiveBigIntegerField()
+        )
+    )
+
+    LOGGER.info(f"Handled {total_count} historical employer summer vouchers:")
+    LOGGER.info(f"- Converted numeric serial numbers to integers: {updated_count}")
+    LOGGER.info(f"- Left as NULL: {total_count - updated_count}")
-import logging
-
-from django.db.models import PositiveBigIntegerField
-from django.db.models.functions import Cast, Trim
-from stdnum.fi.hetu import is_valid as is_valid_finnish_social_security_number
-
-LOGGER = logging.getLogger(__name__)
-
-
-def set_current_valid_serial_number_based_foreign_keys(
-    employer_summer_voucher_model, youth_summer_voucher_model
-):
-    """
-    Convert valid serial number strings in _obsolete_unclean_serial_number to actual valid
-    ForeignKey references in youth_summer_voucher_id in EmployerSummerVoucher model.
-
-    Matching is done first by purely numeric serial numbers, and if that fails, by matching
-    social security number and creation year.
-    """
-    # Real data in production was at start of 2026 around ~8000 EmployerSummerVoucher objects,
-    # so performance should be negligible even without bulk updates or better Django ORM use.
-    total_count = employer_summer_voucher_model.objects.count()
-    matched_by_ssn_count = matched_by_serial_count = failure_count = 0
-
-    for employer_voucher in employer_summer_voucher_model.objects.all().iterator(
-        chunk_size=1000
-    ):
-        serial_number = employer_voucher._obsolete_unclean_serial_number.strip()
-
-        # Purely numeric serial numbers? These should be the majority
-        if (
-            serial_number.isdigit()
-            and (
-                youth_vouchers := youth_summer_voucher_model.objects.filter(
-                    summer_voucher_serial_number=int(serial_number)
-                )
-            ).count()
-            == 1
-        ):
-            youth_voucher = youth_vouchers.first()
-            employer_voucher.youth_summer_voucher = youth_voucher
-            employer_voucher.save()
-            # Successfully matched by numeric serial number, skip to next
-            matched_by_serial_count += 1
-            LOGGER.info(
-                f"Matched employer voucher {employer_voucher.id} to "
-                f"youth voucher {youth_voucher.id} by summer voucher serial number."
-            )
-        elif is_valid_finnish_social_security_number(
-            employer_voucher.employee_ssn
-        ) and (
-            (
-                youth_vouchers := youth_summer_voucher_model.objects.filter(
-                    youth_application__social_security_number=employer_voucher.employee_ssn,
-                    youth_application__created_at__year=employer_voucher.created_at.year,
-                    youth_application__created_at__lte=employer_voucher.created_at,
-                )
-            ).count()
-            == 1
-        ):
-            youth_voucher = youth_vouchers.first()
-            employer_voucher.youth_summer_voucher = youth_voucher
-            employer_voucher.save()
-            matched_by_ssn_count += 1
-            LOGGER.info(
-                f"Matched employer voucher {employer_voucher.id} to "
-                f"youth voucher {youth_voucher.id} by social security number & year."
-            )
-        else:
-            LOGGER.warning(
-                f"Unable to find youth voucher for employer voucher {employer_voucher.id}"
-            )
-            failure_count += 1
-
-    # Log summary of results
-    LOGGER.info(f"Handled {total_count} employer summer vouchers:")
-    LOGGER.info(f"- Matched by voucher serial number: {matched_by_serial_count}")
-    LOGGER.info(f"- Matched by social security number & year: {matched_by_ssn_count}")
-    LOGGER.info(f"- Failed to match: {failure_count}")
-
-
-def set_historical_serial_number_based_foreign_keys(
-    historical_employer_summer_voucher_model,
-):
-    """
-    Convert non-negative integer string values in _obsolete_unclean_serial_number to actual integer
-    values in youth_summer_voucher_id in HistoricalEmployerSummerVoucher model.
-
-    Not trying to make the historical records perfect, just doing the minimum obvious conversion.
-    """
-    total_count = historical_employer_summer_voucher_model.objects.count()
-
-    # Real data in production was Jan 2026 around ~55k HistoricalEmployerSummerVoucher rows
-    updated_count = historical_employer_summer_voucher_model.objects.filter(
-        # Non-negative integer strings with possible leading/trailing whitespace:
-        _obsolete_unclean_serial_number__regex=r"^\s*[0-9]+\s*$"
-    ).update(
-        youth_summer_voucher_id=Cast(
-            Trim("_obsolete_unclean_serial_number"), PositiveBigIntegerField()
-        )
-    )
-
-    LOGGER.info(f"Handled {total_count} historical employer summer vouchers:")
-    LOGGER.info(f"- Converted numeric serial numbers to integers: {updated_count}")
-    LOGGER.info(f"- Left as NULL: {total_count - updated_count}")
+import logging
+from collections import defaultdict
+
+from django.db.models import PositiveBigIntegerField
+from django.db.models.functions import Cast, Trim
+from stdnum.fi.hetu import is_valid as is_valid_finnish_social_security_number
+
+LOGGER = logging.getLogger(__name__)
+
+
+def set_current_valid_serial_number_based_foreign_keys(
+    employer_summer_voucher_model, youth_summer_voucher_model
+):
+    """
+    Convert valid serial number strings in _obsolete_unclean_serial_number to actual valid
+    ForeignKey references in youth_summer_voucher_id in EmployerSummerVoucher model.
+
+    Matching is done first by purely numeric serial numbers, and if that fails, by matching
+    social security number and creation year.
+    """
+    
+    # Pre-fetch all youth vouchers to avoid N+1 queries.
+    # We select related youth_application to access SSN and created_at efficiently.
+    youth_vouchers = list(youth_summer_voucher_model.objects.select_related("youth_application").only(
+        "id",
+        "summer_voucher_serial_number",
+        "youth_application__social_security_number",
+        "youth_application__created_at"
+    ))
+
+    # Build lookup maps
+    # Map: serial_number (int) -> Match object or list of objects
+    serial_map = defaultdict(list)
+    # Map: (ssn, year) -> list of YouthSummerVoucher objects
+    ssn_year_map = defaultdict(list)
+
+    for yv in youth_vouchers:
+        if yv.summer_voucher_serial_number:
+            serial_map[yv.summer_voucher_serial_number].append(yv)
+        
+        # Ensure we have the necessary application data for SSN matching
+        if yv.youth_application:
+            ssn = yv.youth_application.social_security_number
+            created_at = yv.youth_application.created_at
+            if ssn and created_at:
+                ssn_year_map[(ssn, created_at.year)].append(yv)
+
+    total_count = employer_summer_voucher_model.objects.count()
+    matched_by_ssn_count = matched_by_serial_count = failure_count = 0
+    updates = []
+
+    # Iterate over employer vouchers
+    for employer_voucher in employer_summer_voucher_model.objects.all():
+        serial_number_str = employer_voucher._obsolete_unclean_serial_number.strip()
+        matched_voucher = None
+
+        # 1. Try matching by Numeric Serial Number
+        if serial_number_str.isdigit():
+            serial_number = int(serial_number_str)
+            candidates = serial_map.get(serial_number, [])
+            if len(candidates) == 1:
+                matched_voucher = candidates[0]
+                matched_by_serial_count += 1
+        
+        # 2. Try matching by SSN + Year (if no serial match yet)
+        if not matched_voucher and is_valid_finnish_social_security_number(employer_voucher.employee_ssn):
+            candidates = ssn_year_map.get((employer_voucher.employee_ssn, employer_voucher.created_at.year), [])
+            # Filter candidates: must be created BEFORE the employer voucher
+            valid_candidates = [
+                yv for yv in candidates 
+                if yv.youth_application.created_at <= employer_voucher.created_at
+            ]
+
+            if len(valid_candidates) == 1:
+                matched_voucher = valid_candidates[0]
+                matched_by_ssn_count += 1
+
+        if matched_voucher:
+            employer_voucher.youth_summer_voucher = matched_voucher
+            updates.append(employer_voucher)
+        else:
+            LOGGER.warning(
+                f"Unable to find youth voucher for employer voucher {employer_voucher.id}"
+            )
+            failure_count += 1
+
+    # Perform Bulk Update
+    if updates:
+        employer_summer_voucher_model.objects.bulk_update(updates, ["youth_summer_voucher"])
+
+    LOGGER.info(f"Handled {total_count} employer summer vouchers:")
+    LOGGER.info(f"- Matched by voucher serial number: {matched_by_serial_count}")
+    LOGGER.info(f"- Matched by social security number & year: {matched_by_ssn_count}")
+    LOGGER.info(f"- Failed to match: {failure_count}")
+
+
+def set_historical_serial_number_based_foreign_keys(
+    historical_employer_summer_voucher_model,
+):
+    """
+    Convert non-negative integer string values in _obsolete_unclean_serial_number to actual integer
+    values in youth_summer_voucher_id in HistoricalEmployerSummerVoucher model.
+    """
+    total_count = historical_employer_summer_voucher_model.objects.count()
+
+    updated_count = historical_employer_summer_voucher_model.objects.filter(
+        _obsolete_unclean_serial_number__regex=r"^\s*[0-9]+\s*$"
+    ).update(
+        youth_summer_voucher_id=Cast(
+            Trim("_obsolete_unclean_serial_number"), PositiveBigIntegerField()
+        )
+    )
+
+    LOGGER.info(f"Handled {total_count} historical employer summer vouchers:")
+    LOGGER.info(f"- Converted numeric serial numbers to integers: {updated_count}")
+    LOGGER.info(f"- Left as NULL: {total_count - updated_count}")
-import logging
-
-from django.db.models import PositiveBigIntegerField
-from django.db.models.functions import Cast, Trim
-from stdnum.fi.hetu import is_valid as is_valid_finnish_social_security_number
-
-LOGGER = logging.getLogger(__name__)
-
-
-def set_current_valid_serial_number_based_foreign_keys(
-    employer_summer_voucher_model, youth_summer_voucher_model
-):
-    """
-    Convert valid serial number strings in _obsolete_unclean_serial_number to actual valid
-    ForeignKey references in youth_summer_voucher_id in EmployerSummerVoucher model.
-
-    Matching is done first by purely numeric serial numbers, and if that fails, by matching
-    social security number and creation year.
-    """
-    # Real data in production was at start of 2026 around ~8000 EmployerSummerVoucher objects,
-    # so performance should be negligible even without bulk updates or better Django ORM use.
-    total_count = employer_summer_voucher_model.objects.count()
-    matched_by_ssn_count = matched_by_serial_count = failure_count = 0
-
-    for employer_voucher in employer_summer_voucher_model.objects.all().iterator(
-        chunk_size=1000
-    ):
-        serial_number = employer_voucher._obsolete_unclean_serial_number.strip()
-
-        # Purely numeric serial numbers? These should be the majority
-        if (
-            serial_number.isdigit()
-            and (
-                youth_vouchers := youth_summer_voucher_model.objects.filter(
-                    summer_voucher_serial_number=int(serial_number)
-                )
-            ).count()
-            == 1
-        ):
-            youth_voucher = youth_vouchers.first()
-            employer_voucher.youth_summer_voucher = youth_voucher
-            employer_voucher.save()
-            # Successfully matched by numeric serial number, skip to next
-            matched_by_serial_count += 1
-            LOGGER.info(
-                f"Matched employer voucher {employer_voucher.id} to "
-                f"youth voucher {youth_voucher.id} by summer voucher serial number."
-            )
-        elif is_valid_finnish_social_security_number(
-            employer_voucher.employee_ssn
-        ) and (
-            (
-                youth_vouchers := youth_summer_voucher_model.objects.filter(
-                    youth_application__social_security_number=employer_voucher.employee_ssn,
-                    youth_application__created_at__year=employer_voucher.created_at.year,
-                    youth_application__created_at__lte=employer_voucher.created_at,
-                )
-            ).count()
-            == 1
-        ):
-            youth_voucher = youth_vouchers.first()
-            employer_voucher.youth_summer_voucher = youth_voucher
-            employer_voucher.save()
-            matched_by_ssn_count += 1
-            LOGGER.info(
-                f"Matched employer voucher {employer_voucher.id} to "
-                f"youth voucher {youth_voucher.id} by social security number & year."
-            )
-        else:
-            LOGGER.warning(
-                f"Unable to find youth voucher for employer voucher {employer_voucher.id}"
-            )
-            failure_count += 1
-
-    # Log summary of results
-    LOGGER.info(f"Handled {total_count} employer summer vouchers:")
-    LOGGER.info(f"- Matched by voucher serial number: {matched_by_serial_count}")
-    LOGGER.info(f"- Matched by social security number & year: {matched_by_ssn_count}")
-    LOGGER.info(f"- Failed to match: {failure_count}")
-
-
-def set_historical_serial_number_based_foreign_keys(
-    historical_employer_summer_voucher_model,
-):
-    """
-    Convert non-negative integer string values in _obsolete_unclean_serial_number to actual integer
-    values in youth_summer_voucher_id in HistoricalEmployerSummerVoucher model.
-
-    Not trying to make the historical records perfect, just doing the minimum obvious conversion.
-    """
-    total_count = historical_employer_summer_voucher_model.objects.count()
-
-    # Real data in production was Jan 2026 around ~55k HistoricalEmployerSummerVoucher rows
-    updated_count = historical_employer_summer_voucher_model.objects.filter(
-        # Non-negative integer strings with possible leading/trailing whitespace:
-        _obsolete_unclean_serial_number__regex=r"^\s*[0-9]+\s*$"
-    ).update(
-        youth_summer_voucher_id=Cast(
-            Trim("_obsolete_unclean_serial_number"), PositiveBigIntegerField()
-        )
-    )
-
-    LOGGER.info(f"Handled {total_count} historical employer summer vouchers:")
-    LOGGER.info(f"- Converted numeric serial numbers to integers: {updated_count}")
-    LOGGER.info(f"- Left as NULL: {total_count - updated_count}")
+import logging
+from collections import defaultdict
+
+from django.db.models import PositiveBigIntegerField
+from django.db.models.functions import Cast, Trim
+from stdnum.fi.hetu import is_valid as is_valid_finnish_social_security_number
+
+LOGGER = logging.getLogger(__name__)
+
+
+def set_current_valid_serial_number_based_foreign_keys(
+    employer_summer_voucher_model, youth_summer_voucher_model
+):
+    """
+    Convert valid serial number strings in _obsolete_unclean_serial_number to actual valid
+    ForeignKey references in youth_summer_voucher_id in EmployerSummerVoucher model.
+
+    Matching is done first by purely numeric serial numbers, and if that fails, by matching
+    social security number and creation year.
+    """
+    
+    # Pre-fetch all youth vouchers to avoid N+1 queries.
+    # We select related youth_application to access SSN and created_at efficiently.
+    youth_vouchers = list(youth_summer_voucher_model.objects.select_related("youth_application").only(
+        "id",
+        "summer_voucher_serial_number",
+        "youth_application__social_security_number",
+        "youth_application__created_at"
+    ))
+
+    # Build lookup maps
+    # Map: serial_number (int) -> Match object or list of objects
+    serial_map = defaultdict(list)
+    # Map: (ssn, year) -> list of YouthSummerVoucher objects
+    ssn_year_map = defaultdict(list)
+
+    for yv in youth_vouchers:
+        if yv.summer_voucher_serial_number:
+            serial_map[yv.summer_voucher_serial_number].append(yv)
+        
+        # Ensure we have the necessary application data for SSN matching
+        if yv.youth_application:
+            ssn = yv.youth_application.social_security_number
+            created_at = yv.youth_application.created_at
+            if ssn and created_at:
+                ssn_year_map[(ssn, created_at.year)].append(yv)
+
+    total_count = employer_summer_voucher_model.objects.count()
+    matched_by_ssn_count = matched_by_serial_count = failure_count = 0
+    updates = []
+
+    # Iterate over employer vouchers
+    for employer_voucher in employer_summer_voucher_model.objects.all():
+        serial_number_str = employer_voucher._obsolete_unclean_serial_number.strip()
+        matched_voucher = None
+
+        # 1. Try matching by Numeric Serial Number
+        if serial_number_str.isdigit():
+            serial_number = int(serial_number_str)
+            candidates = serial_map.get(serial_number, [])
+            if len(candidates) == 1:
+                matched_voucher = candidates[0]
+                matched_by_serial_count += 1
+        
+        # 2. Try matching by SSN + Year (if no serial match yet)
+        if not matched_voucher and is_valid_finnish_social_security_number(employer_voucher.employee_ssn):
+            candidates = ssn_year_map.get((employer_voucher.employee_ssn, employer_voucher.created_at.year), [])
+            # Filter candidates: must be created BEFORE the employer voucher
+            valid_candidates = [
+                yv for yv in candidates 
+                if yv.youth_application.created_at <= employer_voucher.created_at
+            ]
+
+            if len(valid_candidates) == 1:
+                matched_voucher = valid_candidates[0]
+                matched_by_ssn_count += 1
+
+        if matched_voucher:
+            employer_voucher.youth_summer_voucher = matched_voucher
+            updates.append(employer_voucher)
+        else:
+            LOGGER.warning(
+                f"Unable to find youth voucher for employer voucher {employer_voucher.id}"
+            )
+            failure_count += 1
+
+    # Perform Bulk Update
+    if updates:
+        employer_summer_voucher_model.objects.bulk_update(updates, ["youth_summer_voucher"])
+
+    LOGGER.info(f"Handled {total_count} employer summer vouchers:")
+    LOGGER.info(f"- Matched by voucher serial number: {matched_by_serial_count}")
+    LOGGER.info(f"- Matched by social security number & year: {matched_by_ssn_count}")
+    LOGGER.info(f"- Failed to match: {failure_count}")
+
+
+def set_historical_serial_number_based_foreign_keys(
+    historical_employer_summer_voucher_model,
+):
+    """
+    Convert non-negative integer string values in _obsolete_unclean_serial_number to actual integer
+    values in youth_summer_voucher_id in HistoricalEmployerSummerVoucher model.
+    """
+    total_count = historical_employer_summer_voucher_model.objects.count()
+
+    updated_count = historical_employer_summer_voucher_model.objects.filter(
+        _obsolete_unclean_serial_number__regex=r"^\s*[0-9]+\s*$"
+    ).update(
+        youth_summer_voucher_id=Cast(
+            Trim("_obsolete_unclean_serial_number"), PositiveBigIntegerField()
+        )
+    )
+
+    LOGGER.info(f"Handled {total_count} historical employer summer vouchers:")
+    LOGGER.info(f"- Converted numeric serial numbers to integers: {updated_count}")
+    LOGGER.info(f"- Left as NULL: {total_count - updated_count}")