March 13, 2020

Better customer service linked to improved local input

Haskayne researcher suggests way to ensure products people want are in stores

Have you ever gone to a store to buy something, only to leave empty-handed because what you wanted wasn’t there?

It has been estimated that one out of every 13 items sought by shoppers isn’t in stock, says Dr. Osman Alp, PhD, who is an associate professor at the Haskayne School of Business. For Walmart, alone, fixing such “stock-outs” may represent a $3-billion U.S. opportunity, he says.

But retailers have found it hard to tackle this problem. Inventory systems typically track products that are in stores, such as how much of a particular item has been sold.

They don’t record the demand for products that currently aren’t there, says Alp, adding customers who don’t find what they want rarely provide feedback. “You don’t tell anybody about this — you just go,” he says. “You may go to another store, you may come back later at another time, or you may not come back at all.”

Problem affected Amazon

As one of the co-authors of a recent paper published in Manufacturing and Services Operations Management, Alp studied the problem as it affects grocery chains. Even a company as technologically sophisticated as Amazon ran into difficulties when it purchased Whole Foods in 2017 for $13.7 billion U.S., he says.

Haskayne researcher, Dr. Osman Alp, PhD co-authored a recent study that explored how stock-outs affect grocery chains.

Haskayne researcher Osman Alp co-authored a study that explored how stock-outs affect grocery chains

Mark Skogen

Although a computer-aided ordering system had earlier been implemented by Whole Foods to centralize inventory decisions, the grocery chain experienced substantial stock-outs and customer dissatisfaction, says Alp.

Part of the problem was that local managers were no longer able to provide knowledge critical to the success of their stores, creating an “information asymmetry,” he says. “Store managers have prompt knowledge of the variations in their local demand, but headquarters can only rely on historical sales data, which doesn’t reflect the amount of lost sales at the local domain.” 

Incentive missing for managers

Under such arrangements, there isn’t much reason for managers — who are often on fixed salaries — to stick their necks out with headquarters by offering feedback, he says. They might be rewarded with things such as extra pay if they meet headquarter revenue targets, “but there is this incentive misalignment because this won’t help reduce stock-outs.

“Revenue represents a very blended figure that is contributed to by several different factors. It doesn’t really reflect headquarters’ objective to maximize customer service by avoiding stock-outs.”

Blindly ordering as many items as possible in case customers want them isn’t the answer, he says. “Keeping larger inventories is very costly, particularly for perishable food, and those costs will get passed on to consumers. Stores must strike a balance between carrying inventories and incurring shortages.”

His paper instead proposes to reduce stock-outs by giving local store managers more power. The solution was successfully tested using a mathematical model that used real data from a European grocery chain, he says.

Local decisions empowered

“Headquarters would still want to have a centralized ordering system to keep track of everything, and to make recommendations about how much of each product to order,” says Alp. But local managers would be able to override the system and make their own decisions.

Their success would be evaluated by a new key performance indicator (KPI), which is a target measured over a certain period of time. Although managers would still be on a fixed salary, this new KPI would reward them “if the item is on the shelf at a predetermined time instant, and if the end-of-period inventory level is kept at a reasonable minimum,” says Alp.

If this KPI is implemented with optimized parameters for every store, headquarters will incentivize local managers to order the right quantity that maximizes customer service, but without requiring headquarters to track the exact demand at each store, he says.

“The system will say, ‘Order 1,000 units of this cereal,’ and then the store manager, based on his own information and trying to maximize his performance, will say: ‘I don’t want 1,000 units — 600 is enough,’ or maybe another time, he will order 3,000,” says Alp. “This will eventually minimize shortages and maximize customer service.”