Successful Use of RFID Requires the Right Infrastructure

Not a day goes by that 7-Eleven CIO Keith Morrow doesn’t dream about RFID technology. Like most CIOs in the retail industry, he believes that widespread RFID adoption is a sure thing and that the technology?which some day will enable him to track every single product, from manufacture to checkout (and possibly beyond), that the $33 billion convenience store chain sells?will revolutionize his business. He also knows that the biggest names in global retailing?Carrefour, Gillette, Home Depot, Marks & Spencer, Metro AG, Procter & Gamble, Tesco and Wal-Mart?are all lining up behind it. And with the holiday shopping season looming (a season coming in a stagnant economy that’s been squeezing retailers dry, forcing them to discount merchandise deeply in order to turn it), Morrow can’t help thinking about RFID’s promise of absolute inventory control, consequent cost reductions and increases in margin.

But to realize those savings and profits before the competition, to roll out the thousands of RFID readers and the millions of RFID tags that will give 7-Eleven Superman-like vision into its supply chain, Morrow has had to devote himself to RFID full time. He campaigns internally for top-level support for RFID by demonstrating the wonders of the technology at the tech fairs he mounts annually for his business colleagues. And for the past two years, he has been conducting pilots to figure out the cost of a full-fledged RFID implementation, its potential ROI and the impact of the technology on 7-Eleven’s existing systems.

Because, as he and everyone else knows, the cost will be high and the impact will be huge.

RFID’s Challenge

RFID technology is going to generate mountains of data about the location of pallets, cases, cartons, totes and individual products in the supply chain. It’s going to produce oceans of information about when and where merchandise is manufactured, picked, packed and shipped. It’s going to create rivers of numbers telling retailers about the expiration dates of their perishable items?numbers that will have to be stored, transmitted in real-time and shared with warehouse management, inventory management, financial and other enterprise systems. Applications of RFID technology are also going to need to rely on a new kind of computing architecture known as edge computing, in which vast amounts of processing will take place at the edges of the enterprise’s network rather than in corporate data centers.

RFID, experts agree, is a transformational technology.

“If this goes all the way to the product level,” says Morrow, “RFID will require levels of bandwidth and access and storage such as we’ve never contemplated.”

And even though a variety of technical, economic and social issues (see “Retailers to Customers,” this page, and “Customers to Retailers,” Page 82) are currently inhibiting widespread RFID adoption, CIOs are keeping close tabs on its development. Whether they’re contemplating implementing RFID technology now, in 18 months or in three years, whether they’re thinking at the pallet, case or item level, they realize now’s the time to begin assessing the changes they will need to make to their IT infrastructures. They know that the future success of RFID will hinge upon how well they prepare the ground now.

“RFID has some prerequisites,” says Bill Homa, CIO of Hannaford Bros., a New England grocery store chain. “People are getting the impression from reading the trade press that RFID solves all your supply chain problems. If you have a pretty good supply chain now, RFID will make it better. If you’re not able to accept advanced shipping notices electronically and match them with your purchase order, RFID isn’t going to help you.”

So executives such as Homa, Morrow, Kathleen Starkoff of Limited Brands, and Saks’s Bill Franks are positioning their companies, through pilots and feasibility studies, to take advantage of the technology, and they’re trying to do so without breaking the bank and without having to overhaul their infrastructures.

RFID Today

The buzz about RFID may make it sound like the new, new thing, but the technology dates back to World War II. According to Bill Allen, marketing manager for Texas Instruments RFid Systems, the technology first was used in the 1940s to identify ships and airplanes as friend or foe. What’s new about it is that it’s becoming cheap enough to put on pallets and cases of merchandise, says John Parkinson, chief technologist for Cap Gemini Ernst & Young. Today, tags cost between 25 and 30 cents, down from 40 cents last year. And as the price falls, more applications are developed.

Even if it’s just applied to pallets, cases and cartons of merchandise, RFID will cut warehouse and distribution costs?some 3 percent to 5 percent of retailers’ revenue?according to management consulting firm Kurt Salmon Associates. Readers placed throughout warehouses will pick up signals from tags without the need for a human to point a scanner at the tag, which is the way it’s currently done with bar codes. This will enable retailers to reduce warehouse and distribution center staff.

RFID’s fans also claim that it will help solve some of the most complex and costly problems in retailing, including loss and theft of merchandise (which Parkinson says cost the industry an estimated $30 billion a year) and out-of-stocks, which cost grocery stores as much as 4 percent of their revenue. Armed with more accurate information about what’s on their shelves, in their stock rooms and on its way from factories or distribution centers, retailers will be able to make out-of-stocks the exception rather than the rule, and that in turn will enable them to sell more, satisfy demand, improve service and increase inventory turns.

For example, with tags on cases of wine that transmit information about their location, bottlers and retailers will know whether their products have arrived at their destinations. If a case of Chianti gets picked off a truck by a hijacker or swiped from a distribution center or simply lost, retailers will know approximately when and where it happened and will be able to hold truck drivers and distribution center managers accountable. “Smart” shelves and stores outfitted with readers will know exactly where merchandise is located throughout the store and will alert staff when shelves need to be restocked or merchandise needs to be reordered?before customers pick shelves clean.

Indeed, the ROI on these kinds of RFID applications is so promising that Wal-Mart, the 800-pound wholesaling gorilla, issued a ukase last July demanding that its top 100 suppliers put RFID tags on all their pallets, cases, cartons and high-margin items by January 2005. (The Department of Defense issued a similar announcement last October, requiring its suppliers to tag all pallets and cases by 2005.) Action on the RFID front has grown much more intense for retail CIOs since Wal-Mart made this announcement. (See “The Wal-Mart Factor,” Page 86.)

Truly universal adoption and implementation of RFID technology depend on the cost of tags dropping to a nickel, a level analysts believe will be achieved in three to five years. Once that happens, it will be possible (and profitable) to tag every pack of Dentyne and every box of Ronzoni Rigatoni. Until then, item-level tagging is more realistic for high-margin items, such as consumer electronics, jewelry and haute couture. A Best Buy or a Gucci can easily justify the cost of putting a 30 cent tag on a $1,000 sound system or $250 silk shirt if that tag will help track lost merchandise and prevent theft.

While RFID may present a compelling value proposition to retailers that sell high-ticket merchandise or that wish to tag at the case, carton or pallet level, there are technical obstacles to widespread adoption that still need to be overcome. For example, the radio waves by which tags communicate are absorbed by liquids and distorted by metal, making RFID useless for tracking, say, cans of orange juice. (The Auto-ID Center at MIT is working to fix these problems.)

More important, to prepare their stores for RFID, CIOs need to take a long, hard look at their IT infrastructures.

RFID Storage:An Avalanche of Data

Limited Brands CTO and Group Vice President Starkoff believes the influx of data that RFID will generate will force CIOs to rethink their data warehousing strategies, much as she’s doing at her company. CIOs will have to get smarter about what they store and how they store it. They’ll have to measure the data’s ROI and decide what should be trashed and what should be saved based on the data’s age and the cost of storing and retrieving it. “We’ll get rid of some data that we’ve been storing that maybe doesn’t have the return,” says Starkoff.

She adds that the determination of what data to retain will be tied directly to business processes. For example, once an item is sold, a company may want to retain data indicating that fact, but it may want to purge the RFID-generated supply chain data about the item’s journey from the factory plant to the store.

Starkoff also thinks that CIOs may find relief from the data glut through software being developed by Manhattan Associates, SAP and other vendors that collects the reader data and turns it into clear and concise messages that say that the shipment is as it should be or that there is an exception to the order.

To mitigate the impact of all this data on Limited Brands’ network, Starkoff plans to leverage the network management tools the company currently uses to optimize the bandwidth she has today. Also, because the tags allow storage of data specific to each unit?whether pallet, case or garment?it’s possible for readers to determine the content of a box or the origin of a particular shipment without having to make a network connection to a central database. She says this will minimize bandwidth demands.

Starkoff realizes that if Limited Brands has to spend for bandwidth and storage upgrades on top of its investment in tags and readers, the total cost of an RFID implementation will skyrocket.

“If you deploy RFID technology without thinking it through and without optimizing your infrastructure, you can see where the ROI would just dissipate,” she says. So, to minimize her investment and mitigate the impact of all this RFID data on Limited Brands’ infrastructure, she’s trying to figure out what information needs to be transmitted in real-time and what can wait 24 hours for a batch update. Right now, she believes inventory and replenishment information will be most valuable in real-time. “When RFID moves beyond the supply chain and onto the sales floor, real-time RFID information could make for a dynamic, sales-driven replenishment system,” she says.

RFID Remediation: Y2K Redux

Today, most retailers’ systems have been written to hold 11-digit UPC bar codes. But the serial numbers encoded on RFID tags, known as electronic product codes (EPCs), are composed of 13 digits. To accommodate those two extra digits, CIOs are going to have to expand the numerical structures inside their systems.

An initiative called Sunrise 2005, launched in 1997 by the Uniform Code Council (UCC), a standards body for the retail and manufacturing industries, mandates that U.S. and Canadian companies be capable of scanning and processing up to 14-digit bar codes by Jan. 1, 2005. If they don’t, they won’t be able to share information with their trading partners and they’ll experience time-to-market delays and added costs. According to the UCC, this will require retail CIOs to either replace legacy systems with so-called RFID-ready systems or undertake a Y2K-like effort to reprogram their systems so that they recognize 13-digit EPCs.

Retailers such as Saks Fifth Avenue that sell a lot of European-manufactured merchandise have a head start on this database expansion because the serial numbers on European products contain 13 digits. Because Saks Executive Vice President and CIO Franks started preparing for Sunrise 2005 a few years ago and has already remediated many of his systems, he’s got a good foundation for RFID.

Franks inventoried all of Saks’s systems and all of the documents Saks exchanges electronically with its 1,900 suppliers to ensure that they contain enough space for global trade item numbers and EPCs. If they didn’t, those systems had to be remediated in much the same way that Saks readied its systems for Y2K. He says that now he’s going over all the information from purchase order to advanced shipping notice to invoice to payment acknowledgement to make sure Saks is in compliance with Sunrise 2005 and therefore will be able to support RFID when the company pilots it in early 2005.

“We will approach the enhancements on a priority basis as the business value and operational necessity dictates,” says Franks, none too cheerfully.

RFID Integration: Forward to the Back End

Two years ago, 7-Eleven piloted a VIP (Virtual Instant Payment) card at the 7-Eleven store inside the company’s corporate headquarters in Dallas and at a store in Plano, Texas. The VIP card was equipped with an RFID chip and functioned like a debit or credit card. By waving her VIP card near an RFID reader by the cash register, a customer could pay for purchases without having to stand in line or fumble through her purse or swipe her debit card through a reader (at just the right speed)?much as ExxonMobil customers use an RFID-enabled SpeedPass to pay for gasoline. This was the company’s first foray into RFID, and it helped CIO Morrow understand what system changes he needed to make in order to integrate this new application with the company’s back-end systems.

One of Morrow’s biggest challenges was getting information about sales conducted with the VIP card into the company’s primary store system, its Retail Information System, which provides the in-store tools for point of sale, scanning, ordering, merchandising, receiving and various other management reporting functions. “We couldn’t have a standalone transaction network and then not clear that against our product sales,” says Morrow. So he had to create a service that ran on the back-office computer and pulled transactions made with VIP cards into 7-Eleven’s cash reports. He also had to make sure that the store system had a way to identify VIP card transactions, in much the same way that purchases made using cash, checks, credit cards or debit cards were identified as such. To that end, he programmed a new payment mechanism for the VIP card into the system so that when a customer wants to pay with a VIP card, a sales clerk can select “VIP card” from the point-of-sale system instead of cash, debit, credit or check.

“The integration work was laborious,” Morrow recalls. “[But] the actual hardware pieces?once we got them tuned to read at the distance we wanted, which was two to three inches?were pretty straightforward to deploy.”

Brett Kinsella, general manager of the supply chain management group for IT consultancy Sapient, says the type of RFID application will determine which enterprise system it will get hooked up to. For example, RFID at the pallet and case level will have to be linked to warehouse management systems. RFID at the item level will touch virtually every system from inventory management to replenishment to CRM systems. But because item-level tagging is still a ways off, few if any companies have given this level of integration?particularly with CRM systems?much thought. CIOs who have implemented an infrastructure for enterprise application integration are going to have the easiest time moving RFID data into existing enterprise systems, says Kinsella, “because they can look at this data as another node on their EAI servers.”

RFID Architecture: Computing on the Edge

As companies deploy RFID technology, their enterprise architectures will become more distributed, says Sanjay Sarma, cofounder of the Auto-ID Center at MIT, the consortium developing the infrastructure to enable supply chain applications of RFID technology. Rather than using a data center at corporate headquarters to aggregate and process the avalanche of data that will be generated by items, cartons and cases being read throughout supply chains, that work will be done at the edges of the corporate network?on store shelves, at the point of sale, at loading docks and on forklifts. Sarma calls this distribution of intelligence and computing across a wider network “edge computing.” The distributed architecture, he says, will help companies manage the data glut.

“If you take all that data [from RFID] and supply it to the home office, bandwidth explodes,” says Sarma. “But if you distribute it outward, you reduce the bandwidth you need back to the home office. You reduce computing in the home office, and you make things [operate] much faster.”

Part of this architecture?and the key to enabling RFID applications for retailers and manufacturers?are two technologies that Auto-ID, a partnership between nearly 100 global companies and five research universities, is developing: Savant and object name service (ONS). RFID readers will be wired into a computer system running Savant, an application that manages all the data going in and out of readers. The ONS matches the EPC to the address of a server, which contains information about the product. By calling up an IP address, the ONS essentially tells you what the EPC means, working in a similar manner to the way webpages are called up on the Internet.

Sarma doesn’t think the move toward edge computing will require CIOs to rethink their IT architectures fundamentally. He says CIOs simply will have to invest in more localized computing systems. “Done right, this may not change the central infrastructure. You may create more intelligence at the edges of the network, which will make operations more efficient,” he adds.

And that’s the driving force behind RFID: to decrease retailers’ costs and increase their margins by keeping better track of inventory and enhancing service to customers.