Closing the Gap on Stockouts

July 18, 2010

Andres Musalem's new model can help retailers and brand managers mitigate lost sales.

What happens when shoppers find empty spaces on store shelves instead of their preferred brands of shampoo, soda or other products?  Do they buy different brands, or remain loyal and head to another store in search of their favorite products? And can a price reduction convince shoppers to buy the product again once it is back on shelves?

These questions have challenged brand managers and store owners for ages.  Sophisticated inventory tracking systems can help reduce product outages, known as stockouts, in some retail settings.  However, these systems are not available in all retail settings, and even the best systems don't completely eliminate stockouts.

Duke University professor Andres Musalem and co-authors have developed a new model to help managers estimate the effects of stockouts and find the best ways to recapture lost sales. 

"It can be difficult for managers to know how long their products have been out of stock, and thus how many sales they may have lost," Musalem said. "Our model can help managers overcome this lack of information and identify the best ways to mitigate lost sales."

The team tested the model using real-world data from shampoo sales of six supermarkets in Spain. Their findings, published in the July issue of Management Science, demonstrate the value of using mathematical models and simulation methods to understand consumer responses to stockouts and to design plans to protect a retailer or a manufacturer's revenue under situations of insufficient inventory.

In the case of shampoo purchases in Spain, Musalem found customers were likely to purchase another product in the same store when fewer than five types of products were missing from the shelf.  But when more than five products were out of stock, stores sacrificed 20% of their expected normal sales.

"A store manager could use this analysis to balance product availability with staff workloads and inventory costs," Musalem said.  "In the case of our shampoo study, we learned that you can tolerate a certain amount of empty spaces on your shelves before restocking, but you have to be careful not to let the product selection slip to the point where sales drop precipitously."

The model also demonstrated that discounting similar products can discourage customers from heading to another store to find their preferred product. In the shampoo example, the model indicated discounting a similar item is a good way to overcome a stockout of one particular product, but it's better to discount an already popular product if many of a store's brands are out of stock.

"Implementing an analysis of this type can help brand managers and retailers protect their revenues and customer loyalty," Musalem said.

More information about the model, including the researchers' full paper, "Structural Estimation of the Effect of Out-of-Stocks" is available here.

Contact Info

For more information contact our media relations team at media-relations@fuqua.duke.edu.