Wal-Mart’s Texas Two-Step

At a Wal-Mart Supercenter in Denton, Texas, a hallway leading to the retail floor has two pairs of gray and black, 6-foot-high silhouettes with yellow eye-like lights within 10 feet of each other. The mission: Detect and record goods that contain tags emitting radio waves.

Down the road at a Hickory Creek store, printers from Hewlett-Packard sit on shelves with radio frequency identification (RFID) tags attached. A brief note informs customers that Wal-Mart is testing electronic tagging. The tags sit on the bottom side panel of the printer box. Each tag carries the serial number, model and an undisclosed “option” code as a backup for the bar code.

Welcome to the epicenter of Wal-Mart’s early efforts to track the movement of 21 tagged products wirelessly. Ultimately, all of the 100,000 types of products found in the average Supercenter will be tracked. Here in Texas, Wal-Mart is receiving 21 tagged products from eight suppliers at Distribution Center 6086 in Sanger and distributing to stores in the Dallas-Fort Worth metroplex. Wal-Mart says it will begin bringing more suppliers into the pilot each week, leading up to its January 2005 deadline to have its top 100 suppliers shipping RFID-tagged goods.

Wal-Mart says the RFID trials are a success with the occasional “pilot learnings.” But analysts wonder what constitutes a success for Wal-Mart’s RFID efforts. Is success 100% of Wal-Mart’s top suppliers sending 100% of their goods tracked by radio waves? If so, the RFID mandate is falling short. Is success moving an industry forward toward RFID gradually? If so, Wal-Mart is successful. Does success mean Wal-Mart’s suppliers get a return on investment from RFID? No one knows.

The latter question is a stickler, says AMR Research analyst Kara Romanow, who adds few suppliers will generate a return beyond keeping Wal-Mart happy. “A lot of this is hype,” says Romanow. “[Suppliers] are barely closer to 100% compliance with Wal-Mart’s deadline than [they] were six months ago.”

A trip to Wal-Mart’s Texas beachhead shows it is forging ahead. RFID readers and tagged cases and pallets are arriving from eight suppliers—Gillette, Hewlett-Packard, Johnson & Johnson, Kimberly-Clark, Kraft Foods, Nestle Purina PetCare, Procter & Gamble and Unilever. These tagged cases are tracked by radio waves as they arrive from Sanger and move to the store floor, but the technology is still immature. Wal-Mart wouldn’t discuss its read rate–the ability of a reader to recognize the radio waves from a tag–but tracking tests at Sun Microsystems’ nearby Carrollton facility show the chances of a reader recognizing a tag may be little better than a coin flip. A lot of trial and error is needed to get 100% read rates in the lab, and those results are no guarantee they’ll hold in the real world. That’s what Wal-Mart is wrestling with now.

Wal-Mart has kept a tight lid on the specific details of its pilot, but has acknowledged “a few minor bumps in the road.” Simon Langford, Wal-Mart’s manager for RFID, says all top 100 suppliers will hit Wal-Mart’s deadline of January 2005, and 37 more volunteers will join them. The catch? Not every product will be tagged. “It’s not viable to tag everything,” says Langford.

The reasons why suppliers won’t have 100% of goods tagged vary, according to Langford. Less-than-perfect read rates, lack of returns for suppliers and expense of the tags are all valid reasons to fail to tag all products. Langford says Wal-Mart is willing to work with a supplier that tags 80% of its products and “provide feedback” to get its partner to 100%. It’s too early to discuss fines, he says.

At its supplier meeting last month, Wal-Mart outlined its next set of deadlines. By June 2005, Wal-Mart will be live with RFID-tagged goods in six new distribution centers as well as 250 stores and Sam’s Club outlets. By October 2005, Wal-Mart will install the technology in up to 13 distribution centers and 600 stores. By January 2006, Wal-Mart’s next 200 top suppliers will begin tagging cases and pallets.

According to Wal-Mart, testing has gone as expected, with the retailer evaluating processes such as routing products on loading docks, putting goods on conveyor belts and handling items with faulty tags.

Langford says a pilot works out these issues and insists read rates will hit the 100% mark. “A lot of it is choosing the right tag to get performance,” he says, adding that items such as a case of soup with liquid and metal may need larger antennas on the tags. To work through these issues, Wal-Mart gets tagged cases and pallets of lotion, cat food, shampoo, laundry detergent and peanuts, to name a few.

In the meantime, meeting Wal-Mart’s mandate is going to take a lot of trial and error. Kevin Ellison, manager for Sun Microsystems’ RFID test center in Carrollton, is overseeing testing for two companies-Poly-America, a privately held company in Grand Prairie, Texas, that makes private-label trash bags for Wal-Mart, and tire giant Goodyear. In the Poly-America test, a tagged pallet containing cases of tall kitchen bags was read easily. But the chances of the 140 trash-bag cases on the pallet being read were 53%. The test, which was in the early stages, was hurt by the location of the tags on the boxes and the plastic wrap encasing the pallet.

“It’s not perfect technology and it’s early,” says Ellison. “We obviously don’t have the right tag or have it optimized.”

Ellison has a few guesses why this test failed. The plastic wrap may have shielded the radio waves. The tags were facing down, away from the reader. And the tags could be wrong. There are dozens of tag makers and each antenna design works differently, depending on the properties of the product. A tag from Alien Technology may work on a case of shampoo, but you may need a tag from label printer Avery Dennison for a case of peanuts.

Goodyear’s tires have yielded better results. In the lab, Ellison has gotten 100% read rates on the tires, through weeks of trial and error. Tires, which are tagged individually and often rolled off trucks, present a host of challenges. For starters, their ingredients include oil and steel, two substances that thwart radio waves. Ellison is currently getting 100% read rates by putting tags on the outside of the tire, secured by blue masking tape, and running them through two sets of readers from Alien Technology and Matrics to read two classes of tags. Bottom line: You may need two sets of readers to get 100% read rates.

“If you want 100% read rates, it may make sense to have multiple readers,” says Ellison. Wal-Mart may have the same idea. One hallway in its Denton store has two sets of Matrics readers just about 10 feet apart from each other. The Matrics readers are designed to read both Class 0 and Class 1 tags required by Wal-Mart.

Ellison’s tire challenges are far from over. He will have to find glue that won’t interfere with radio waves. Goodyear will also need tags that won’t fall off when thrown onto a conveyor moving at Wal-Mart’s preferred speed—nearly two football fields a minute.

“I think we’ll see 100% read rates on pallets,” says Ellison. “But cases will be more difficult. How fast they are read will depend on reader speed, where the tag sits and your middleware.”

Speed is no small issue, considering Wal-Mart isn’t going to slow down its conveyors, which move at 600 feet per minute, for RFID. Wal-Mart says it has had some issues, but wouldn’t elaborate. Will Wal-Mart have to change its processes to ensure RFID tags face up or out to a reader?

Some suppliers may have it easier than others. Navin Chandaria, CEO of Conros, the North York, Ontario-based maker of Pine Mountain fire logs, is optimistic. Conros tagged pallets in one plant and plans to roll out RFID systems in another seven. Chandaria hired Hewlett-Packard’s services unit to set up pilots tracking fire logs, which work well with radio waves, unlike liquid, metal and electronics.

Chandaria says he’s not sure about the immediate benefits, but hopes to manage inventory on Wal-Mart’s shelves better. Currently, Conros’ replenishment process isn’t automated. When it’s cold, the company ships a lot of fire logs and hopes stores don’t run out.

He says RFID tracking data with weather pattern information could make better inventory predictions. For now, the data sits in Conros’ Microsoft SQL databases for evaluation.

“If it’s a cold night and there are not enough fire logs on the shelf, it’s not like people come back the next day and buy two logs,” he says.