From Just-In-Time to Real Time

RFID: From Just-In-Time to Real Time

Five years ago, Kevin Ashton, then an assistant brand manager at Procter & Gamble in London, set out to revolutionize the world’s supply chains using a technology called Radio Frequency Identification, or RFID for short. Ashton was frustrated because his team had come up with a hot-selling shade of lipstick that would sell out in an hour, yet not be replaced for a week. When he discovered a smart credit card with a chip that could be read without being swiped through a machine, he hit on the idea of using the same kind of chip in P&G products, to speed corporate response time to the marketplace. Since then, Ashton has sought to make a lot of people believers in the vision he had that these chips, connected through a global network, could enable companies to respond to changes in supply and demand—as well to trouble in the pipeline—in real time. “This is bigger than the Internet,” he says.

Click to download a printable PDF graphic.

At first, people at P&G thought Ashton was nuts. RFID tags—tiny radio-powered microchips that broadcast a unique serial number, like a talking bar code—cost several bucks apiece in those days, and the idea of putting one on every box of cereal and bar of soap was preposterous: Profit margins on such products were measured in cents. But Ashton managed to convince higher-ups at P&G that it was an idea worth exploring. He then got MIT and some of the world’s biggest companies—including Wal-Mart Stores Inc., The Coca-Cola Co., Johnson & Johnson, PepsiCo Inc., The Home Depot Inc. and Unilever—to back his quixotic quest. And now, from an MIT laboratory in Cambridge, Mass., he’s creating converts. Says Dick Cantwell, vice president of global business management for The Gillette Co., a backer of Ashton’s RFID lab, called the Auto-ID Center: “We see RFID as the supply chain technology of the future. It’s going to revolutionize the way we track goods from manufacturing to the consumer and even through recycling.” Ashton puts it another way. “Creating a way for companies to use sensors to identify goods anywhere in the world is a very big deal,” he says. “We are, in effect, creating an Internet of things.”

Ashton essentially sees the world as two giant networks. First, there’s the network of goods and services that make up the global supply chain. Second, there’s the network of computers called the Internet. Ashton wants the machines that make up the world’s supply chains to interact with the Net. “Our vision is a world where by moving atoms you can move bits, and by moving bits you can move atoms,” Ashton says.

But sensors aren’t new, and neither is RFID. General Electric Co. has been using them for years to help its engine customers detect problems before they cause hazards to users. But thanks to new manufacturing processes, recent wireless technology and smarter software, the tiny devices are now at the forefront of an important new trend throughout the manufacturing and consumer goods industries that is already beginning to accelerate the speed at which companies can respond to the marketplace. As with microprocessors and lasers in earlier decades, the novelty is not that these sensors exist at all, but that they have suddenly become cheap enough to be used in ordinary, everyday products—and are already starting to save companies a bundle (see “Making Waves” sidebar). Whether tiny thermometers, mini-microphones, electronic noses, location detectors, motion sensors or RFID tags, all of them can provide information about the physical world and represent a whole new level of automation that experts predict will accelerate the pace of corporate activity, if not change the face of the economy itself.

Before RFID tags, for example, Associated Food Stores Inc. used to have 127 people spend at least part of their day typing information about truck locations into a yard management system. That was not only inefficent; it was inaccurate. The information in the database was wrong 40 percent to 70 percent of the time. Now each truck contains an RFID sensor that automatically signals its location in the yard in real-time for a savings to the company of more than $1 million. The tags made it possible for AFS to get smarter about tracking, and cut 53 trucks and some drivers from the fleet roster. The tags also helped to reduce food spoilage caused by goods sitting in trucks that couldn’t be found right away, for additional savings.

Multiply these speed and efficiency bursts by thousands of potential applications—and new ones, such as the ability of, say, an RFID sensor to track a box of Cheerios from factory to kitchen shelf—and you’re opening up whole new possibilities for everything from inventory control to customer service to customized product delivery.

Days to Hours

Think of it this way: The development of RFID tags is, in some ways, an extension of the decades-old “just-in-time” inventory drive by firms to shrink multibillion-dollar stockpiles into smaller, cheaper ones. In the 1970s, the Japanese undercut U.S. car manufacturing’s production costs by hundreds of dollars per vehicle, just by replacing parts’ storehouses with systems aimed at delivering parts to assembly lines as they were needed. With just-in-time manufacturing, companies cut inventories of parts from weeks to days. Sensors, backers say, will represent a move from days to hours. “Once all of these machines start talking to one another, they’re going to make commerce—and the world—move much faster, more efficiently and at speeds that humans alone couldn’t match,” says Paul Saffo, director of the Institute for the Future, a West Coast think tank.

Sound too good to be true? Think again. Ashton and his researchers have gotten this new “sensor-chain” to work, for the most part, in recent tests. In October, they wired a P&G factory in Cape Girardeau, Mo., with readers and slapped RF tags on pallets of Bounty paper towels. They also wired forklifts at one of Wal-Mart’s Sam’s Club stores in Tulsa, Okla. The goal was to prove that the Internet-like RFID infrastructure designed by the Auto-ID Center could, in the real world, record the movement of goods from the factory to the store with no human intervention. Consider it a test of the first phase of the Internet—but instead of proving an e-mail could be sent over vast distances, Ashton and his team wanted to show that a pallet of paper towels could communicate its movement directly to computers.

As the paper towels left the factory, readers stuck on bay doors with Velcro picked up the unique serial number. A day later, the pallet arrived on a truck at the Sam’s Club in Tulsa, Okla. A reader on the forklift immediately picked up the serial number. A beep on a laptop in the warehouse manager’s office confirmed the arrival of the pallet. The normally understated Ashton calls it “the ping heard ’round the world.”

It wasn’t just that readers had successfully read the tag on the pallet. What was unique in this experiment was that there was an infrastructure that made it possible. The reader passed the tag’s serial number to software that uses the Net to read electronic product codes through an object name service that works something like the Net’s domain name service. The object name service points to a database with information written in physical markup language, similar to XML, that describes the product—and, put simply, tells the computers at Sam’s Club that a pallet of paper towels just arrived.

Not everything worked like a charm. In the Sam’s Club test, the store’s existing bar code scanners “collided” with the frequencies emitted by the RFID tags, causing mixed signals and some devices not to respond. In addition, flawed hardware sometimes kept the RFID readers from picking up all sensors. In both cases, Ashton says, he and engineers are designing better readers and tags—and are also trying to fine-tune the algorithms so RFID tags don’t suffer interruptions in signal strength from competing frequencies.

The test has two more parts. In February, the Auto-ID Center cranked up the information load on the system by tagging cases of P&G’s Pantene Pro-V shampoo and Gillette’s Mach3 razors. And in a final phase of the test, scheduled for this summer, the team will put new low-cost tags, priced at 10 cents each, on individual boxes of cereal, soft drink bottles and cans of Coke in an effort to see if it is possible to track individual items from the factory to store shelves—and see if companies throughout the supply chain can use sensors to respond to real-time demand signals—rather than basing production on forecasts that are invariably wrong. “With RFID, the potential exists to never have to do physical counts of inventory again,” says Larry Kellam, P&G’s director of business-to-business supply chain innovation. Kellam says RFID could cut P&G’s total inventory in half, while simultaneously reducing the number of times a product is not on the shelf. So-called “out-of-stocks” occur about 7 percent of the time, on average, in the packaged goods industry, according to a study by PricewaterhouseCoopers. If the trillion-dollar industry could cut that rate in half, that would free up some $70 billion in savings for other investments.

Another big goal for all manufacturers is to cut supply chain costs. Kevin O’Marah, vice president of supply chain strategies at AMR Research, says losses due to stolen, damaged, misdirected or lost shipments amount to 3 percent to 5 percent of total supply chain costs. “Supply chain inefficiencies are still a gigantic pool of potential value to go after,” says O’Marah. “If RFID makes companies even 1 percent more efficient, you’re talking about a huge amount of money.” P&G’s Kellam believes his company could save some $400 million annually if RFID lives up to expectations and reduces inventory by 50 percent.

Of course, there are a lot of skeptics—and with good reason. RFID has been around since the invention of radar during World War II and has been seen as a promising technology since the 1960s, when the U.S. Department of Agriculture implanted RFID sensors under the skins of cows to track their movements. The Pentagon is using RFID to keep track of military supplies sent to Afghanistan.

Still, it is the cost of each tag—and some technical glitches—that have been keeping sensors out of the commercial mainstream. Even now, some memory-packed tags can cost as much as $5 each. Simpler, less powerful chips are cheaper, but price depends heavily on volume. Buy 1,000 chips and they may cost $1 each. Buy 1 million and each may cost 50 cents. P&G produces more than 20 billion items a year. Putting a 50-cent tag on every item would cost the company $10 billion a year. RFID backers, therefore, have sought to push the price of each tag to less than 5 cents.

They’re getting close. In the past few months, for example, Ashton’s Auto-ID Center has designed a chip with the minimum components needed to carry a 96-bit electronic product code. Alien Technology, a start-up in Morgan Hill, Calif., will be the first to manufacture RF tags based on the spec. If companies backing Ashton’s research buy tags in bulk from Alien, the price of tags could fall to three and a half cents by 2005, Ashton says. But cost isn’t the only headache. The lack of standards governing the use of RFID is also a problem. Companies don’t want to invest in a system if their tags can’t be read by trucking companies, retailers and other businesses. MIT is focused not just on developing low-cost tags and readers but also on formulating global standards to ensure widespread use.

Says Jack Sparn, CIO of CHEP International of Orlando, Fla., “The problem is the entire supply chain hasn’t bought into it yet,” he says. “If you’re the first person with a mobile phone, it’s useless. But as soon as everyone has one, it becomes valuable.”

Critical Mass

Consider the railroad industry. About 10 years ago, all the major lines in North America agreed on a standard for RFID tags and readers. The industry as a whole spent $200 million to install 3,000 readers, and tag 1.5 million cars and 20,000 locomotives. It has paid off. Fort Worth, Texas-based Burlington Northern Santa Fe Corp. used to employ an army of some 500 clerks armed with pencils and clipboards to walk up and down the tracks at its depots and switching stations to read numbers painted on the sides of its railway cars. The information was then handed to data entry personnel who would key it in, so the company’s mainframe systems could track cars. Today, all of the company’s rail cars are tagged (about $30 per tag) and it has 443 readers ($50,000 apiece, including installation) positioned at key junctures along 33,500 miles of track in 28 states and two Canadian provinces. As a result, the company eliminated all of its trackside clerks in 1997. Now, the system has paid for itself, and it reads 100,000 tags a day with virtually no errors. Burlington can provide customers with more accurate data about where their shipments are. The system also has dramatically reduced track delays. “In the old days, when a car was out of place, people had to spend hours trying to figure out where it should be, and that would cause delays throughout the system,” says Shannon McGovern, Burlington’s director of network support systems. “Today, that’s almost never a problem.”

Bigger Role for CIOs

If RFID catches on across industries, the technology will heap new duties onto the shoulders of CIOs, Ashton says, namely, the need to help companies start managing an enormous amount of new information. “The biggest challenge is deciding what to track, where to track it, and link that back into your systems,” says Simon Ellis, supply chain futurist for Unilever. “That’s where companies will find the competitive advantage. It won’t be in the RFID technology, because that will be a commodity.”

CIOs will also have to work with business managers to fine-tune systems. Since the whole point of RFID is to take people out of the loop, software will have to capture business knowledge and apply it automatically. At the most basic level, IT departments will have to work with store managers and suppliers to determine optimal inventory levels, when to trigger an order for more goods, and when to drop the price of slow-moving items. “The CIO won’t take over the decision-making,” says AMR’s O’Marah. “They’re going to be the ones who architect and deliver the infrastructure that allows these automated processes to be built into the way a business runs.”

The trend also has implications for marketing. Tags placed in manufactured items or packaging could spark opportunities to develop new products and enhance existing ones. Unilever’s Digital Futures Laboratory in Englewood Cliffs, N.J., showcases a kitchen that keeps a real-time inventory of foodstuffs and displays a list of “in-stock” items. Invensys, a British maker of industrial equipment, is working with appliance manufacturers such as Whirlpool Corp. to develop ovens that cook turkeys based on instructions from chips in the packaging.

Alien Technology recently won a $120 million contract from the Department of Defense to combine RFID tags with other types of sensors to pick up vibrations or detect the presence of chemicals or biological agents. The U.S. military wants to drop so-called “smart dust” sensors on a battlefield, and by picking up vibrations and knowing the exact location of a specific tag, generals could know how many enemies are hiding in a location or whether chemical or biological weapons are being stored there. RFID tags may even be combined with tiny microphones that look like seed burrs that could attach themselves to someone’s socks, so the military could listen in on conversations.

All this, of course, raises privacy concerns. Chris Hoofnagel of the Washington-based Electronic Privacy Information Center, a nonprofit pro-privacy group, says: “RFID tags, themselves, are passive, but there are going to be any number of entities who will want to use the information collected to track individuals or groups. The issue is control: Can you determine when the tag is active and who is using the information collected? That’s going to be the challenge.”

No question there. But Ashton says sensors can also be used to boost security. One day, Ashton says, RFID sensors may be put in food containers to ensure that products haven’t spoiled or been tampered with. If the sensors become cheap enough, Alcoa CSI, which makes 50 billion bottle caps a year, wants to put them in its products.

Will RFID live up to its promise? If companies like Wal-Mart, P&G and Unilever begin deploying RFID systems next year, they could encourage suppliers to adopt the technology and, Ashton predicts, the network effect would force more companies to get on board. “RFID will either take off quickly,” says Ashton, “or it won’t take off at all.”

Given the benefits companies like Ford and Burlington Northern are getting from RFID so far, the smart money appears to be on smarter machines.

MARK ROBERTI is founder and editor of RFID Journal, an independent Web site that covers the development of RFID technology. He has written about supply chain technologies for The Industry Standard, Business 2.0 and other publications. Comments on this story can be sent to

Making Waves

Radio frequency sensors offer companies more information about the physical world— for savings in time and money.

Ford Motor Co.

Before: Assembly-line workers running low on parts would have to pick up a phone and call the replenishment department to get more, then wait—sometimes for hours.

Now: Ford puts RFID tags on each parts bin. Warehouse operators now know, in seconds, when supplies run low, and automatically deliver parts as needed to workers on the assembly line.

Procter & Gamble Co.

Before: P&G used bar codes to track shipments of goods from factory to retail outlets, but couldn’t do much to halt sudden supply shortages on store shelves.

Now: RFID is tracking shipments, and, eventually, individual products, so they can be stocked on demand in stores. P&G expects to cut its $3.5 billion inventory in half and cut costs by $400 million a year.

San Francisco Int’l.

Before: Security staff at SFO had to hand-carry passenger bags flagged by an FAA profiling system to a bomb detection machine, then back to the check-in counter, causing delays.

Now: RFID tags are put on suspect bags and routed automatically to bomb detection devices, enabling SFO to cut bomb detections security spending by an estimated 50 percent.

Top 5 Fastest-Growing RFID Applications

1. Point-of-sale tracking

2. Rental item tracking

3. Baggage handling

4. Real-time location systems

5. Supply chain management




RFID Handbook

By Klaus Finkenzeller with Rachel Waddinsky

John Wiley & Sons Inc., 1999

Understanding Smart Sensors, Second Edition

By Randy Frank

Artech House Inc., 2000

Web Sites

This newsletter for the RFID industry is produced by AIM, the global trade association for the automatic identification data capture industry.

Copyright © 2002 Ziff Davis Media Inc. All Rights Reserved. Originally appearing in CIO Insight.