Everyone in the retail industry stopped and took notice when Wal-Mart Stores declared in June that it will urge its top 100 suppliers to deliver pallets and cases equipped with radio frequency identification (RFID) tags by 2005. Any directive issued by the world's largest retailer has the potential to drive sweeping adoption, and this particular one could spell major changes for supply chain management.
Wal-Mart thinks that the nascent technology, which can automatically identify a container's contents without requiring line-of-sight scanners, can help to reduce the costs associated with tracking inventory.
Given that Wal-Mart moved 2.5 billion cases through its distribution centers during one six-month period last year, it's not hard to imagine the savings that the company might realize by reducing the time and labor associated with inventory tracking.
One of the chief suppliers to the retail industry, Procter & Gamble, has another angle. The Cincinnati-based company estimates that 10 percent to 16 percent of its products may be out of stock at any moment. Reducing that number by even 10 percent or 20 percent could mean a revenue boost of between 1 percent and 3 percent, says Larry Kellam, director of business-to-business supply network innovation at the consumer goods maker. With over US$40 billion in annual revenue, that would translate to more than $400 million in new revenue.
But neither suppliers nor retailers will realize much benefit until the technology overcomes a series of technical and engineering hurdles. For instance, the tags need to come down in price. To do that, manufacturers need orders for billions of tags, and they need to improve their manufacturing processes to support those volumes.
Tag readers also need to improve in both performance and price. In addition, the software infrastructure to handle RFID tag data must advance past the work-in-progress stage, and standards need to be established to enable different vendors' tags and readers to work together using a wide range of radio frequencies.
"It's one of the most overhyped technologies that we're talking about today," says Jeff Woods, an analyst at Gartner. "It's going to require a lot of creative thinking and hard work to get from vision to reality.
Readers Are Fundamental
A tag reader communicates with an RFID tag and passes on the information to software applications. Readers work with passive tags using a method known as inductive coupling, in which a reader's antenna creates a magnetic field with the tag's antenna.
Although passive tags keep costs down, their readers can still cost more than $1,000, and most read only chips that use a single frequency. To address the problem, the Auto-ID Center designed reference specifications for software-based "agile" readers that can read different types of tags and tags that operate at varying frequencies. Ashton predicts that the reader cost can be cut to $100 to $200 each at a volume of 10,000 units within three years.
"That's an important step, because it means you no longer have to have a proprietary tag-reader combination," says Jim Crawford, vice president of Retail Forward, a research and consulting firm in Columbus, Ohio. "It lets you put in a single infrastructure to read multiple tags."
But Paula Rosenblum, an analyst at AMR Research, says there's little evidence that the price of readers is dropping. Many readers don't work reliably, she adds.
The read range of a tag depends on the the reader's power, the frequency that the reader and tag use to communicate, and antenna size.
Gene Alvarez, an analyst at Meta Group, says a powerful reader will be needed to read a passive tag from long distances. The read range for a passive tag is now three to six meters, he says.
The Software Conundrum
Even if all the tag and reader issues are worked out, simply slapping tags on pallets, cases or individual products and installing readers won't produce the real-time flow of data that retailers and suppliers need to gain the full benefits of RFID technology. RFID is going to change business processes so fundamentally that users will have to either install new, possibly experimental applications or endure a massive rewrite of existing programs, warns Gartner's Jeff Woods.
"I don't see anything (happening) with RFID-centric warehouse management or manufacturing, or even retail processes," he says. "It's a classic innovators' dilemma, because everyone is so heavily invested in bar-code-based infrastructure and processes that they are the least likely ones to make the wholesale transition quickly."
The first applications will emerge in the next two to three years, Woods says. Emerging vendors are working on the problem, as are established vendors such as SAP and IBM.
"Through 2007, we're going to see primarily applications that use RFID tags in the context of bar-code-based processes -- things like receiving at the back door with an RFID tag instead of a bar code," Woods predicts. "It's the three-to-seven-year time frame when we will start to see entirely new processes come about."
The Auto-ID Center's response to managing the flow of data is special-purpose server software, called a Savant, which it predicts will be running in stores, distribution centers, offices and factories. Savants will gather, store and act on information and interact with other Savants, deciding which information needs to be forwarded up or down the supply chain, the center claims.
Under the Auto-ID Center's proposal, RFID tags will contain a limited amount of information in a 64- or 96-bit electronic product code (EPC). The reader pulls the EPC from the tag and passes it to a Savant, which in turn forwards it to an Object Name Service server and then a Physical Markup Language server on a local network or the Internet to find information stored about the product. The Savant can then retrieve the file and forward it to the company's inventory or supply chain applications.
"The Auto-ID Center moved the problem of data from the tag into the system," says Steve Halliday, president of High Tech Aid in Gibsonia, Pa. But he predicts that some companies will want tags that can store more data so they can find out the contents of pallets and cases on the spot where the tags are scanned, rather than having to connect to a Savant and other servers.
"All that RFID does for a retailer or a manufacturer is give them more granular information about their products," says Crawford. "Mastering that process is the critical efficiency issue for the next 20 years easily for retailers and manufacturers."
Tags Get Cheaper
An RFID tag, also known as a transponder, contains an antenna and a microchip that transmits information about the tagged item to a reader. The tag reader then converts the radio waves returned from the tag into a digital form that can be passed to computer systems.
The technology has been used for years to track animals, collect tolls on highways and grant access to buildings. But cost has kept RFID tags from being used on a large scale to identify and track goods in the retail supply chain.
P&G's Larry Kellam says tags were a dollar apiece in 1999 when the company began looking at RFID technology to curb counterfeiting and retail theft and reduce out-of-stock situations. So P&G joined The Gillette Co. and Uniform Code Council as founding sponsors of the Auto-ID Center, an industry-funded research project at MIT.
One of the Auto-ID Center's chief missions has been to find a way to reduce the cost of RFID tags. The center recommends the use of passive tags containing a limited amount of information, because chips with less memory are cheaper. Passive tags draw power from electromagnetic waves that the tag's readers generate, whereas more expensive battery-powered active tags broadcast signals.
The Auto-ID Center's researchers also realized that the tag's silicon chips would need to be smaller to lower the cost. But reducing the size of the chips isn't easy, since robots have trouble handling chips that are the size of pieces of glitter, notes Kevin Ashton, executive director of the Auto-ID Center, who is on loan from P&G.
Alien Technology, an Auto-ID Center sponsor, is at the forefront of a new chip-packaging process called fluidic self-assembly that it hopes will reduce the cost of passive tags from 50 cents in small quantities today to 5 cents at a volume of 10 billion by 2006, says Tom Pounds, vice president of corporate development and product strategy. An Alien manufacturing line capable of producing a billion units annually will go online early next year, and a second manufacturing line capable of producing 10 billion units per year is planned for 2005, Pounds says.
Gillette made waves earlier this year when it negotiated a deal with Alien to purchase up to 500 million tags. Company spokesman Paul Fox says Gillette will achieve its goal of a sub-10-cent tag for field tests over the next few years, although Gartner's Jeff Woods says he thinks Alien is losing money on that deal.
Even though Gillette is doing pilots with European retailers on individual items, the company doesn't foresee item-level tagging in production for at least 10 years, according to Fox. For that to happen, per-tag costs must drop to a penny or less, he says.
Both Wal-Mart and Gillette have decided to focus on pallets and cases. The retailer this spring scrapped plans for a store trial with Gillette. Wal-Mart spokesman Tom Williams says item-level RFID is years away for the retailer.
Those testing the technology are encountering challenges at the pallet and case level. Woods says some users are experiencing read rates of less than 80 percent with tags. Wal-Mart found radio-wave interference problems in field tests of 500 pallets of paper towels.
And metal and liquids don't mix well with radio waves. That meant that P&G had to test different tags, since Bounty towels have different properties than Pantene shampoo.
"These problems are not fundamental," Ashton claims. "They will be solved with time and experience."