Good forecasting prevents empty supermarket shelves
July 6, 2015 | 2 min read
Predicting the right stock levels is complex. The effects of a heat wave on people's purchasing behavior are very different than those of a day of hot weather, says Academic Director Freek Aertsen of TIAS Master of Operations and Supply Chain Excellence in this column.
Image: © Nationale Beeldbank
Last week, such supermarket chains as Albert Heijn and Jumbo had to deal with the problem of empty shelves. The hot weather and, in the case of Albert Heijn, the “Hamster Weeks” promotion campaign created a strong demand for a number of items. Although supermarkets say that they take the weather forecast into consideration when stocking, the demand for certain products, such as ice cream and beer, is larger than expected.
Predicting the right stock levels is a challenge. The data entered strongly determine the end figures. The weather is, of course, one of the variables. But even then, people's purchasing behavior is very different during a heat wave than on a hot day. Albert Heijn's stocks have also been affected because of the “Hamster Weeks.”
All companies struggle with this. My company Eyeon will cooperate with a weather bureau to collect data about the weather. We can use that data to create forecast models. We want to figure out the right way of forecasting. Is it really true that in the past companies always sold more beer when the weather got warmer? If so, then how much more? What is the effect of “Hamster Weeks”? You use all these data, historical data and future data, to build a model.
Using these forecast models is becoming increasingly important as supply chains get leaner. Companies get more and more stock from the chain. This means that it becomes increasingly sensitive to disruptions.
This also increases the effect of a bad forecast. How fast can you remedy an empty stock? A supermarket is dependent on suppliers. Do those suppliers have enough stock? This might not be a problem with Coca Cola, but it is with perishables. And then there is the issue of delivery. Supermarkets have a limited number of refrigerated trucks, which causes longer product delivery times. Because of this, there is always a delay before supermarkets can refill the shelves.
But you still need to enter the right data in these models. A good model does not mean that the right amounts are on the shelves. My colleague lives in The Hague, where it was 24 degrees Centigrade when it was 35 degrees here in Eindhoven. The Tour de France cycling race started in Utrecht. A company can predict that it will sell on average 10 percent more beer because of the hot weather. However, if you increase stocks by 10 percent at one location but the temperature there does not rise by that much while more beer is needed elsewhere, supermarkets will still have empty shelves.
Would you like to make more accurate predictions so that your organization will not end up with empty shelves? Forecasting is discussed in the TIAS Master of Operations and Supply Chain Excellence program.
More information about MOS