Cut raw material waste by 23% for a Shropshire food producer
We introduced simple AI forecasting tools that aligned ingredient orders with historical weekly bakery demand, reducing spoiled stock.
We helped Oakfield Bakery Ltd in Shropshire fix their ingredient ordering system. By connecting their weekly sales data to a simple automated demand calculator, they cut dough and filling waste by 23.4% in twelve weeks. We do not do fancy slides. We configure software that works.
The challenge
Oakfield Bakery Ltd operates a 1,240-square-meter production floor in Telford, Shropshire. Every Tuesday morning, the production planner had to estimate how many flour sacks, butter blocks, and fruit fillings to order for the upcoming week. They relied on handwritten clipboards and manual spreadsheets. This process took nearly six hours of tedious double-checking.
Because retail demand from regional supermarkets shifted based on local weather and school holidays, the planner often over-ordered ingredients. In Q3 2023, the bakery threw away an average of 342 kilograms of fresh ingredients every single week. This spoiled stock was costing them £1,415 per week in lost margins, and their disposal fees were rising by 11% annually. They needed a practical way to match ingredient orders with actual weekly shop floor demands without hiring another full-time manager.
Our approach
Our team of two engineers from Birmingham spent three days on the Telford shop floor. First, we mapped out the path of ingredients from the delivery bay to the main mixing bowls. We quickly realised that historical sales data was trapped in an old Sage accounting system, while production logs lived on paper sheets. We needed to bridge this gap. (By the way, we always look at the physical shop floor first, not just the databases.)
Instead of suggesting expensive enterprise platforms, we built a lightweight data pipeline. We extracted 17 months of weekly sales logs from Sage and cleaned the dataset. Our lead analyst, Sarah Jenkins, then set up a basic statistical forecasting model that weights recent orders against seasonal patterns. We tested this model on past data from Easter 2023 to verify its accuracy before writing a single line of production code.
The solution
We built a simple dashboard using Python and SQL that runs on a cheap tablet at the receiving desk. The system automatically pulls weekly sales data every Thursday at 16:00. It then calculates the exact quantity of raw materials needed for the following Monday's shifts. The planner no longer has to guess how much flour or yeast to order.
From clipboards to cloud databases, we kept the user interface incredibly simple. The tablet displays three clear columns: ingredient name, calculated demand, and recommended order size. The planner can override any recommendation with a single tap if they know a specific local event is coming up. We also integrated automated email alerts that notify suppliers directly once the planner approves the order, saving another three hours of phone calls every week.
Results
Within twelve weeks of launching the system, Oakfield Bakery Ltd reduced their raw material waste from 342 kilograms to 262 kilograms per week, a drop of 23.4%. They saved £3,825 in raw material costs during the very first month of full operation. The purchasing team now spends only 45 minutes on weekly orders instead of six hours.
Timeline
-
January 2024Three-day shop floor audit in Telford and extraction of Sage sales logs.
-
February 2024Built Python statistical model and tested against 2023 supermarket demand history.
-
March 2024Installed receiving-desk tablet and trained two production planners on the interface.
-
April 2024Fully automated supplier emails and handed over system maintenance.
"We used to throw away bins of perfectly good butter and fruit fillings every Friday. This simple tablet system took the guesswork out of our buying. Our weekly order prep now takes under an hour."