Dillard’s, U. of Ark. Study Quantifies RFID’s Superiority to Manual Inventory Counts

By Claire Swedberg

The results, based on a project carried out in three of the retailer's stores, involved more than 1,000 pairs of men's jeans labeled with EPC Gen 2 RFID tags.


The RFID Research Center at the University of Arkansas’ Information Technology Research Institute (ITRI) has released a research paper indicating that not only does radio frequency identification improve inventory accuracy, it can also take the place of annual, manual inventory counts. The paper describes the results of the third phase of a pilot studying item-level RFID technology in the retail environment.

The study focused on three Southeastern stores operated by Dillard’s, one of the United States’ largest apparel retailers, with more than 300 stores in 29 states. The results are based on tests that took place in fall 2008—coincidentally, during one of the stores’ prescheduled twice-yearly inventory counts. Bill Hardgrave, executive director of the ITRI and director of the RFID Research Center, presented the study’s results at RFID Journal LIVE! 2009, held last month in Orlando, Fla.

Bill Hardgrave

Although Dillard’s had scheduled a manual inventory at all of its stores, the University of Arkansas’ research team had not anticipated that to occur in the midst of the study. “We had already started the study,” Hardgrave says, when the inventory process began. “We decided that could make the results even more interesting.”

The study was conducted at three of the retailer’s area stores. Initially, four sites were involved—two test stores, each issued fixed and handheld RFID interrogators, as well as two control stores—but one control store was closed during the study. During the research project, a manufacturer of men’s denim jeans attached hangtags with EPC Gen 2 RFID inlays to each of its garments, then shipped them to the three stores. In the midst of the study, the retailer conducted its semiannual inventory counts at all three locations, and the study then continued, tracking the benefits of RFID after that semiannual inventory brought all three stores to near-perfect levels of inventory accuracy.

What the researchers found was significant, Hardgrave says. Following the semiannual inventory, the accuracy of the control store’s inventory quickly began to decline, while the two RFID test stores improved upon the accuracy established by the semiannual count.

As a result of the study, Hardgrave notes, researchers determined that not only does RFID raise inventory accuracy, it can actually eliminate the need for the disruptive, expensive and time-consuming annual or semiannual inventory check most retailers conduct to improve inventory accuracy, as well as for reporting to accountants or the IRS.

“Results clearly indicate the improvement in inventory accuracy due to RFID,” Hardgrave states. But the study also found that RFID counts could replace manual, scheduled storewide counts. “What this study shows is there is no need to do those inventory counts,” he says, since RFID can be just as accurate, if not more so.

The study was the third phase of a retail pilot launched by Dillard’s and the University of Arkansas. Phases one and two studied the use of RFID on apparel and footwear in a simulated store. In those first two studies, researchers examined the hardware’s read ranges, read rates and effectiveness (see New Research Papers from the University of Arkansas). The third phase was intended not only to test the technology in a real-world environment, but also to determine the business value of RFID.

The newest study focused on whether the technology could reduce two common challenges for retailers—overstock and understock of inventory. In both cases, whether the store is missing inventory that should be on its shelves, or has more product than it should have, the store suffers expense. According to multiple studies, theft, cashier errors and shipping problems, in addition to counting mistakes, result in inaccurate inventory counts in more than 50 percent of stores.

With the third phase of the pilot—conducted for 10 weeks, on Mondays and Fridays before the stores opened—a third-party inventory-auditing group performed a count of men’s denim jeans at the control and test stores. To accomplish this, the group employed bar-code scanners to read the bar-coded stock-keeping unit (SKU) label attached to each pair of jeans.

The test stores were equipped with fixed Gen 2 UHF interrogators at the dock doors, to read tagged items (each tag transmitted a unique ID number linked to a product SKU and description in the back-end system) as they arrived from the manufacturer. To track potential theft or store associate error, readers were also installed at employee exit doors, as well as at the customer exit door nearest the department in which the jeans were stocked. At the point of sale (POS), tags were removed from the products and discarded, and inventory was tracked based on the SKU sold. The store collected information from each RFID read, in order to know when product arrived, and when it left the site, either by POS data or RFID reads at the exits. Researchers also utilized handheld RFID readers to take inventory of tagged jeans on shelves located on the sales floor at the two test stores. Between 1,100 and 2,000 items were tagged and counted during the study.

Following the first five weeks of the study, all three stores began their scheduled semiannual storewide inventory count. That process, using bar-code scanners and manual counting, took two weeks to complete, after which the study resumed its use of RFID for three more weeks. The project found that the semiannual manual inventory count improved inventory accuracy by 17 percent, and that in the next three weeks, RFID further improved inventory accuracy by approximately another 4 percent in the two test stores, according to reports by the third-party inventory-auditing group. In the control store, however, inventory accuracy improved by 12 percent following the count, then declined by 13 percent in the three weeks following. In other words, the benefits of the manual inventory had been lost within that period of time in the store that did not use radio frequency identification, while in the test stores, RFID improved inventory accuracy.

With RFID, Hardgrave asserts, based on this research, a thorough storewide inventory could be conducted more often than twice annually—perhaps every week—in a matter of minutes using RFID readers. This, he says, would yield the benefits of storewide inventory without requiring the exhaustive process of manual inventory counting by employees, which often takes weeks.

Researchers also tested the accuracy of RFID as a theft-reduction or security device, by imitating a typical shoplifting scenario—such as removing a stack of jeans, placing them under an arm, and running or walking past the reader out the doors. In nearly all cases, the tags were read, while in several instances, when one tagged item was held closely under an arm, the system failed to capture its ID tag.

Hardgrave’s group also analyzed RFID data to provide loss-prevention insight. Because the sales staff had been instructed to remove a pair of jeans’ RFID tag at the point of sale, any tags detected by the interrogators at the employee or customer entrances would serve as an indicator of potential theft or store associate error (that is, a failure to remove the tag). To determine the latter, the researchers examined each RFID read from the employee or customer entrance, then compared that information with POS data. If a match was discovered, the team attributed the read to a worker’s error. If no sale of that product was found, the item was deemed stolen.

Over the course of the study, more than $3,500 in merchandise was assumed stolen based on the above approach. Dillard’s could then use the resulting data to determine the quantity of items stolen, as well as the time and place (the employee or customer entrance). The retailer could then utilize this information to institute appropriate loss-prevention methods and adjust the inventory count accordingly—and, if necessary, order additional merchandise to replace the stolen goods.

Dillard’s did not respond to a request for comments regarding the study, though the retailer, as well as the University of Arkansas, “will be aggregating these results for an overall picture,” Hardgrave says. “We know we’ve seen solid results.”