diff --git a/erpnext/manufacturing/report/exponential_smoothing_forecasting/test_exponential_smoothing_forecasting.py b/erpnext/manufacturing/report/exponential_smoothing_forecasting/test_exponential_smoothing_forecasting.py
index 3f8ffd74af9..37d6c7da7ab 100644
--- a/erpnext/manufacturing/report/exponential_smoothing_forecasting/test_exponential_smoothing_forecasting.py
+++ b/erpnext/manufacturing/report/exponential_smoothing_forecasting/test_exponential_smoothing_forecasting.py
@@ -8,17 +8,24 @@ from erpnext.manufacturing.report.exponential_smoothing_forecasting.exponential_
 	execute,
 )
 from erpnext.selling.doctype.sales_order.test_sales_order import make_sales_order
+from erpnext.stock.doctype.item.test_item import make_item
 from erpnext.tests.utils import ERPNextTestSuite
 
-TEST_ITEM = "_Test Item"
 FROM_DATE = "2026-06-01"
 TO_DATE = "2026-08-31"
+SMOOTHING_CONSTANT = 0.5
 
 
 class TestExponentialSmoothingForecasting(ERPNextTestSuite):
 	"""Drive real submitted Sales Orders and assert the report buckets the ordered
 	quantities into the correct historical periods and produces a forecast."""
 
+	def setUp(self):
+		# The forecast query has no lower date bound, so it would pick up any committed
+		# Sales Order for the item. A uniquely-named item keeps the buckets scoped to
+		# just this test's orders.
+		self.item = make_item(properties={"is_stock_item": 1}).name
+
 	def test_monthly_qty_forecast_from_sales_orders(self):
 		# Historical demand: distinct calendar months strictly before FROM_DATE.
 		# Monthly period keys are derived from the period's last day (e.g. "mar_2026").
@@ -45,26 +52,29 @@ class TestExponentialSmoothingForecasting(ERPNextTestSuite):
 		for key, qty in history.items():
 			self.assertEqual(flt(row.get(key)), flt(qty), f"bucket {key} mismatch")
 
-		# A forecast is produced for the first future period. The first non-zero
-		# historical period seeds the forecast at the average of non-zero months,
-		# so the future forecast must be positive.
+		# The forecast seeds at the average of the non-zero historical months and then
+		# smooths through them in order: F = F + a*(actual - F). Asserting the exact
+		# analytical value pins the smoothing formula (Jun 2026 works out to ~7.2083).
 		expected_avg = sum(history.values()) / len(history)
-		self.assertGreater(flt(row.get("forecast_jun_2026")), 0.0)
-		self.assertLessEqual(flt(row.get("forecast_jun_2026")), max(history.values()))
 		self.assertAlmostEqual(flt(row.get("avg")), expected_avg, places=6)
 
+		forecast = expected_avg
+		for month in ("mar_2026", "apr_2026", "may_2026"):
+			forecast = forecast + SMOOTHING_CONSTANT * (history[month] - forecast)
+		self.assertAlmostEqual(flt(row.get("forecast_jun_2026")), forecast, places=6)
+
 	def test_ignores_documents_outside_range_and_other_docstatus(self):
 		self.create_sales_orders({"2026-05-10": 6})
 		# A draft SO and a future-dated SO must not contribute to historical demand.
-		make_sales_order(item_code=TEST_ITEM, qty=100, transaction_date="2026-05-20", do_not_submit=True)
-		make_sales_order(item_code=TEST_ITEM, qty=100, transaction_date=FROM_DATE)
+		make_sales_order(item_code=self.item, qty=100, transaction_date="2026-05-20", do_not_submit=True)
+		make_sales_order(item_code=self.item, qty=100, transaction_date=FROM_DATE)
 
 		_columns, row = self.run_report()
 		self.assertEqual(flt(row.get("may_2026")), 6.0)
 
 	def create_sales_orders(self, date_to_qty):
 		for transaction_date, qty in date_to_qty.items():
-			make_sales_order(item_code=TEST_ITEM, qty=qty, transaction_date=transaction_date)
+			make_sales_order(item_code=self.item, qty=qty, transaction_date=transaction_date)
 
 	def run_report(self, **extra):
 		filters = frappe._dict(
@@ -76,16 +86,16 @@ class TestExponentialSmoothingForecasting(ERPNextTestSuite):
 				"periodicity": "Monthly",
 				"from_date": FROM_DATE,
 				"to_date": TO_DATE,
-				"smoothing_constant": 0.5,
-				"item_code": TEST_ITEM,
+				"smoothing_constant": SMOOTHING_CONSTANT,
+				"item_code": self.item,
 			}
 		)
 		filters.update(extra)
 
 		columns, data = execute(filters)[:2]
 		item_row = next(
-			(r for r in data if r.get("item_code") == TEST_ITEM),
+			(r for r in data if r.get("item_code") == self.item),
 			None,
 		)
-		self.assertIsNotNone(item_row, f"{TEST_ITEM} row missing from report output")
+		self.assertIsNotNone(item_row, f"{self.item} row missing from report output")
 		return columns, item_row