Shulman has been managing the retailer's
RFID rollout for approximately a year. "One hundred percent of our RFID-enabled stores outperform our non-RFID stores," she states. "They are doing twice as good—meaning if you were to have an average of 3 percent in comp sales at non-RFID stores, then the average in RFID stores would be 6 percent in comp sales." (Comp sales, or comparable store sales, compare sales achieved during a given period to those from the same span of time during the previous year.)
At its clothing factory in Los Angeles, American Apparel applies hangtags embedded with passive ultrahigh-
frequency (
UHF)
EPC Gen 2 inlays to all items headed for an RFID-enabled store. Once these stores receive the shipments, employees use RFID readers to capture each tag's unique ID number, which is linked to the garment's stock-keeping unit (SKU). The RFID software then reconciles each number against the store's order information, in order to ensure that the shipment is complete.
Based on the information collected, the inventory software then informs the workers regarding which items need to be moved to the sales floor for shelf restocking, and which should be put into storage. Employees then use a fixed
interrogator to
read the RFID tags once more as garments leave the store's back room, to confirm that the items are, indeed, being brought out to the sales floor. (Each RFID-enabled store keeps bifurcated inventory for its back stock and its sales floor, so that at all times, it knows the stock levels in the back stock versus on the sales floor.) If this
reader collects the RFID tag number of any item that should not be brought to the sales floor, or if it fails to collect the tag ID of a garment that is supposed to go to the floor, the software triggers an alert, thereby showing which items are missing or superfluous. Employees also perform regular inventory counts on the sales floor, using handheld readers. This information is compared with the POS data, and the software generates a list of any discrepancies that may exist between the two databases.
"There are about 35,000 items in each store," Shulman says, "and we generally don't have a discrepancy of more than 100 items between the inventory and sales data." Stores must reconcile and account for the differences, which are often just a matter of a delay in the software systems updating the latest data. An item might have been sold, for instance, between the time that the POS data was collected and when that information was compared with the inventory data.
American Apparel has also tested the use of RFID tags for
electronic article surveillance (
EAS). According to Shulman, a pilot project performed last year—in which RFID readers were erected at store exits to test whether the EPC Gen 2 tags attached to items could be used to trigger security alerts when unsold items left the store—failed. The interrogators installed at the exits could not reliably detect the tags, she says, and the company determined that it would simply be too easy for thieves to prevent the tags from being read, by doing things such as holding them in their hands to block them from being read. So instead, she says, American Apparel has deployed a traditional EAS system at select stores, using EAS hard tags that are removed from items at the point of sale. In some locations at which the company has deployed both the
item-level RFID system and the EAS solution, shrinkage has been reduced by as much as 75 percent, she says, attributing a reduction in external theft to the EAS system and a decrease in internal theft to
radio frequency identification.