Background & Targets
Founded in 1792, WHSmith made its name as a bookseller and stationer. Over the years it has diversified its range of products to include food-to-go, such as sandwiches, salads and pastries, and is today one of Britain’s best-known retailers, with a fast growing international business.
Previously, the retailer’s fresh goods supplier was responsible for forecasting, replenishment and spoilage risk. However, WHSmith realized it needed to put an effective forecasting and replenishment solution in place quickly, one that could handle split deliveries and changes in demand due to promotions and holidays. WHSmith also wanted to reduce waste. The company approached RELEX in late 2016 and was impressed by its comprehensive solution offering as well as its strong track record in managing fresh food.
WHSmith has a central replenishment team that manages ordering for all its stores. The goal was to find a system that would be easy for the team to use and that could be implemented quickly. The chosen system would need to handle forecasting and replenishment of chilled products, multiple deliveries a day, and would need to have an excellent track record with fresh food replenishment.
Project
The project’s specification phase began in the first week of December 2016 and proceeded quickly. Initial configuration, and integration with a range of existing systems, was completed by mid-January. The first store was brought onto the system by the end of February and RELEX was rolled out to all 350 UK stores by mid-April.
The project remained on track even though there were a number of changes at WHSmith during the implementation period including a relaunch of its food range.
Using Machine Learning to Forecast Footfall
RELEX uses machine learning abilities to include airport passenger numbers into the forecasts so that WHSmith can use airport traffic figures to predict demand. The system processes data provided by airports and early indications show that this can help improve sales forecast accuracy.
“It’s great to have found a solution provider that is eager to tackle new challenges head on. When we mentioned the opportunity to use airport traffic numbers, the RELEX team was quick to find an innovative way to utilize this data in our demand forecasts,” says Banwait.
“We’re excited as we get to push the boundaries of retail forecasting, and show how our systems can be adapted to customers’ needs,” says Kevin Wales, Head of Client Partnerships UK at RELEX Solutions.
The Results
RELEX provided a system that met the customer’s needs around fresh food by offering optimized weekday profiles and managing several deliveries a day for some products, splitting quantities appropriately between each. The short shelf life of fresh food items makes accurate forecasting and replenishment essential, and as sales may fluctuate day to day, best results are achieved through day-level forecasting and replenishment. Managing split deliveries was also important, as the stores have limited capacity to store items and the products need to be as fresh as possible when they reach the shelves.
The key result is that the 12 weeks pre vs. post implementation showed a significant reduction in spoilage and the expectation is that it will continue to fall. At the same time availability increased.
Jag Banwait, Merchandise Controller, WHSmith
“The key result is that the 12 weeks pre vs. post implementation showed a significant reduction in spoilage and the expectation is that it will continue to fall. At the same time availability increased,” says WHSmith’s Merchandise Controller Jag Banwait. “The RELEX team is very solutions-oriented, quick and flexible in their way of working.”
RELEX’s easy configurability has allowed the WHSmith team to optimize the system and their processes further, especially with regard to safety stocks, and in so doing has gained even closer control of their business. With routine replenishment largely automated, the team is now able to concentrate far more on planning and managing exceptions, areas where its expertize really adds value.