Back to the Present
For our study, we had a food producer as a partner. It produces goods that are not necessarily needed on a daily basis – they stay on the supermarket shelves for a relatively long time and are known as "slow movers". These products, which are rarely sold, make up the majority of the range in most retailers – we customers appreciate the widest possible selection, even if we don't really need most of them on a daily basis.
When will the next order arrive?
Only when the stock of these slow-moving goods falls below a certain level, when enough customers have bought that particular spice or that one type of vinegar, does the supermarket order more. However, this should then be delivered by the manufacturer as quickly as possible. This leads to a tricky situation: depending on demand, the orders only arrive very irregularly. Sometimes there are days, sometimes weeks, sometimes months between individual orders.
At the same time, the product cannot be produced in large quantities in stock, but should ideally be made available shortly before the order is placed so that the products have the longest possible shelf life in the supermarket and at the consumers. Ideally, it would be necessary to have as precise an insight as possible into how much product stock is still available in the supermarket so that production can be started at the right time - even before the actual order is placed. However, this information is not available to most manufacturers.
Nowcasting - forecasting the present
So how can you forecast the current product stock? We have already made use of existing data: the supermarket records exactly how many and which products are sold each day. Our manufacturer can purchase this data for a fee. We now know when our manufacturer last delivered the product to the supermarket, how many products are sold in the supermarket every day and how much time our manufacturer needs from the order being placed to delivery. With these three pieces of information - as our theoretical and empirical studies show - the time of the next order can be calculated much more precisely than with the statistical methods and empirical values that have been used to date. By the way: Nowcasting is also used to estimate the current number of coronavirus cases or the current overall economic development.
In principle, our method can be applied to all products for which daily sales data is available and for which orders are placed on an irregular basis. The advantages are obvious: more precise forecasts allow our manufacturer to optimize its processes, saving time and money. Back to the present - it's worth it.
Full publication: "When is the Next Order? Nowcasting Channel Inventories with Point-of-Sales Data to Forecast the Timing of Retail Orders" accepted for publication at the European Journal of Operational Research (JOURQUAL A, IF 6.4, ABS 4).
Prof. Dr. Kai Hoberg
Kai Hoberg is Professor of Supply Chain and Operations Strategy at Kühne Logistics University since November 2017. Kai Hoberg’s current research topics include supply chain analytics, the role of technology in supply chains, inventory modeling, and the link between operations and finance. In particular, he explores the fundamental drivers of supply chain performance and strategies applying real-world data. His research findings have been published in academic journals like Journal of Operations Management, Production and Operations Management or European Journal of Operational Research. Besides research, Kai Hoberg is very enthusiastic about teaching supply chain management applying new teaching concepts.
Connect with Prof. Kai Hoberg at LinkedIn or learn more here.
You want even more?