Fast Food Logistics
Let's get started: HAVI optimizes and manages its customers' supply chains—end to end.
Who hasn’t done it yet? Almost everyone has eaten at McDonald’s. The burger chain is the most valuable fast-food company in the world—and the number one in this country, too. Its roughly 1,500 German locations generate more than three billion euros in annual revenue. That’s a whole lot of cheeseburgers—it’s the top seller in Germany. What makes it special: a bun, beef, processed cheese, onion, pickle, ketchup, and mustard sauce.
The logistics service provider HAVI ensures that the restaurants always have sufficient stock of these and the other roughly 140 so-called “demand units” required for the individual recipes—and under no circumstances more than necessary. Not too little, not too much: that is HAVI’s mission and promise of success. The “accuracy rate” of its daily sales forecasts is over 90 percent. And this isn’t about crates, boxes, or kilograms: it’s about every single cucumber and tomato.
How does HAVI do it? How does such precise demand planning work?
“It is allocated down to the material level based on the products sold and the underlying recipes,” explains Torsten Oldhues, Managing Director of Operations at HAVI Germany. Of course, that’s much easier said than done. In fact, the required level of accuracy is underpinned by a complex forecasting system. The efficiency of HAVI’s entire supply chain management strategy stands or falls with it. It ensures that product availability is planned for at low inventory levels throughout the entire supply chain—end-to-end, from the respective producers to the stores.
Which product is sold daily, and how often?
“For the forecast, we use multiple linear regression with historical point-of-sale data,” Oldhues continues. At its core, it is based on historical sales figures from the last two to three years. For new locations, data from suitable “reference restaurants” is used. In this way, the forecast predicts anew each day which product will be sold at which location and how often. The forecast then calculates how many of the respective demand units are needed, including the required lettuce leaves, tomato slices, and dollops of sauce. From this, it determines the individual shipping units required for each product.
However, fine-tuning is also necessary for a forecast to work properly, according to Oldhues. “To calculate demand planning, we enter certain causal factors into the planning system, such as weekends, holidays, or local events like festivals, soccer games, markets, and trade shows. Trends and seasonality can be adjusted individually.” Reliable forecasts can even be generated for the approximately 100 special promotions, such as the currywurst that was available at McDonald’s restaurants for six weeks. Here, the company draws on the data pool of reference products, similar to what is done for new store openings—and just as efficiently.
“Special-offer items should still be available on the last day of the promotion until closing time—and then be sold out. Our forecasting process can achieve that,” he continues. “The sales rate through the end of the promotion is 97 percent; the remaining inventory after the promotion ends is one percent.”
Last Blind Spots
According to Oldhues, statistical outliers are essentially generated only by bus travelers. “Tour buses account for the last blind spots in the forecast. Where is the coffee tour headed, at which rest stop will the school trip stop, and when will the tour bus take a break? You could compile statistics on tour buses, but nothing would come of it.” Yet another forecast. You can take Torsten Oldhues’s word for it.
