For 30 years, retail supply chains -- manufacturers, warehouses, delivery systems, and retailers -- have struggled to eliminate running out of stock. Not only does it kill sales for a given item, but studies show it also causes some buyers to delay shopping altogether until the back-ordered item arrives. According to the Grocery Marketing Association, the out-of-stock percentage for all retailers hovers around 7 percent.
Recently, the industry has turned to analytics engines to predict when items might go out of stock, notes TrueDemand Software CEO Eric Peters, but these systems have a couple of flaws.
First, they tend not to deal with the status of individual stores, so they miss when one store’s overstock masks another’s deficit. Second, they don’t link to systems that actually restock items or, even better, prevent out-of-stock events from occurring in the first place. “There’s not a direct relationship between taking retail orders and putting enough product on the shelf,” Peters says.
TrueDemand aims to fill that gap with predictive tools that analyze point-of-sale data, warehouse data, delivery data, and historic purchase trends -- such as time of day and the effect of price and shelf locations -- at individual stores. Rather than make periodic forecasts, TrueDemand’s software continually analyzes the data to keep its predictions current.
IT can also integrate predictions with planning and execution tools, so immediate action can be taken. For example, IT could set policies that automatically change the delivery schedule for stores that habitually run out of an item on the same days. Other issues could be brought to the attention of purchasing staff at the retailer.
“We can tell a consumer-goods company that this particular Wal-Mart has 13 of these items and in three days will be out of stock,” Peters says. And as retailers deploy RFID tags on their products, TrueDemand’s software will be capable of predicting out-of-stock events at specific times during the day.
Later this year, the company plans to release new software that helps analyze the effects of delivery on the labor needed for stock management and delivery. It might be more cost-effective, for example, for a retailer to allow 1 percent out-of-stocks than to have extra deliveries increase labor costs.
TrueDemand attributes its breakthrough predictive abilities to several factors, including powerful commodity computers that can crunch volumes of retail data in real time and an industry shift to joint analysis and planning. But it credits its own team of scientists, most of whom were demand-planning researchers at Stanford University, for developing the algorithms that make the vast quantity of data involved useful.
Perfect stock levels will always be elusive. But TrueDemand promises to tighten the retail supply chain and help ensure that consumers get what they want when they want it.