|
Mix a promising but immature and costly technology with fast-moving adoption mandates from a huge customer to which you can't afford to say no. Throw in a pinch of inevitable human errors, stir in the unyielding laws of physics, and then top it all off with a dash of bickering about standards. And what you get is the passive radio-frequency identification market near the end of 2005.
As Wal-Mart prepares to add 200 more suppliers in January, and as initiatives from outfits ranging from Best Buy to Target to the Defense Department get under way, RFID adopters continue to struggle with problems that have been around the technology for two or more years, and they're finding new ones, too. These challenges won't stop RFID's inevitable forward movement, but problems with data filtering and analysis, clunky tag-read work-arounds, continuing high costs, and unrealized returns on investment are causing temporary derailments. They're also breeding frustration--and maybe even a backlash.
Gillette Co., an early advocate, began shipping about 200 RFID-tagged product types to Wal-Mart's Sanger, Texas, distribution center in March. But Gillette has reduced the count to an undisclosed number, concentrating on fast-selling products like razors and items that do best in promotions, says Dick Cantwell, VP of the global value chain for EPC and retail availability at Gillette.
Wal-Mart's ultimate goals always have been to use RFID across the company's supply chain to speed inventory to store floors and eliminate out-of-stock items. But Gillette may not be the last company to decide that it wants a more-targeted approach. For some products, such as DVDs, manufacturers are exploring RFID-tagging only during the first couple of weeks or months after release, when sales are greatest. Meanwhile, logistics champs FedEx Corp. and UPS Inc. don't yet see a business case for widespread use of RFID, not when bar codes do a fine job tracking packages, for less money.
Annin VP Beard found RFID tag pricing confusing and initially underbudgeted for them.
Companies are seeing progress and pushing RFID technology forward, but it's limited in scale and scope. Consumer-goods maker Kimberly-Clark Corp., an early proponent of the technology, has been moving rapidly ahead with its deployment. The company ships more than 200 items, from diapers to detergent, in RFID-tagged cases and pallets to Wal-Mart, Albertsons, and Target distribution centers in the United States, and overseas to Metro Group and Tesco. Its RFID research lab last month became one of just two in North America accredited as testing centers by the main RFID standards groups for consumer goods and retail. But even Kimberly-Clark still flows RFID data into a separate database rather than integrating that information with its ERP systems. That step will be key to realizing a return on investment by pinpointing shipment whereabouts and out-of-stock goods. "We spent the year learning compliance, physics, and infrastructure requirements to get the performance out of the technology," says Mike O'Shea, director of corporate Auto ID and RFID strategies at Kimberly-Clark. "Now we have to start taking a look at how to deliver the value."
Bicycle maker Pacific Cycle LLC has firsthand knowledge of the integration nightmares that come with trying to merge RFID data into its SAP ERP system. Data-formatting issues and software incompatibilities make it hard to import accurate data, never mind the extra steps needed to gain valuable insights. "We have a heck of a time with the amount of middleware, trying to get one to talk with another," IS director Ed Mathews says. Pacific Cycle has dedicated a person to sift through the 70,000 records it receives weekly from the more than 50 RFID-tagged products it ships to retailers, searching for inconsistencies that crop up time and again. That includes duplicated data, such as multiple reads being generated when a pallet of tagged bicycles gets stalled near a reader at a distribution center.
Given all the headaches, most companies are skating around the difficulty of integrating RFID data into their core manufacturing processes. In the face of painful realities, they aren't even trying to take full advantage of RFID's promise. Spending on RFID systems and services is trending up, with AMR Research predicting $1.9 billion next year going up to $4.2 billion in 2009. But it's the rare consumer-goods company that spends close to earlier predictions of up to $22 million on a deployment, AMR research director Kara Romanow says. Instead, most companies are taking a risk-averse approach, with the average consumer-goods company spending about $662,000 on simple slap-on-a-tag-and-ship deployments. "Rather than spend the millions of dollars we predicted it would take for manufacturers to implement RFID the correct way and integrate it into business processes and applications, they did the bare minimum to comply with retail mandates," she says.
Getting the bare minimum right can be hard enough, especially with the 3-year-old problem of inconsistent tag read rates. Annin & Co. is a 158-year-old flag manufacturer that hopes to comply with Wal-Mart's mandate by November, well ahead of its January 2007 deadline. But wild fluctuations in tag read rates are forcing a makeshift data-collection process at its Ohio warehouse. During trials, pallet read rates of the tag's electronic product code, developed by RFID standards body EPCglobal Inc. to identify items, range from 40% to nearly 100%. Companies need to do a tag read in-house before shipping their products, to verify that the label works and collect tracking data. So Annin set up a conveyer system to scan and collect RFID data one case at a time, before the pallet is assembled. The goal had been to read all the cases on a pallet with one scan. But "the technology isn't there yet," says Carter Beard, Annin's VP of manufacturing and distribution. The one-at-a-time approach provides accurate reads but takes its toll on time and resources.
Many companies have struggled with read rates. Kimberly-Clark early on had to weed out tags that were dead-on-arrival, and "there was too much work up front" to keep weak tags from getting turned into labels, O'Shea says, though the work has paid off with near-perfect read rates. Wal-Mart, too, reads each case individually to provide more accurate EPC data for internal users and suppliers--an extra step that wasn't part of the original plan.
Wal-Mart officials argue that there has been plenty of progress. "We've seen more than 5 million cases go through the participating stores," says Simon Langford, Wal-Mart's RFID strategist. "That equates to about 58 million electronic product code reads." But as Wal-Mart's deployment grows--from 130 consumer-goods companies shipping RFID tags on 65% of their products to two distribution centers, to more than 300 suppliers in January shipping goods to five distribution centers--things could get more cumbersome. Wal-Mart, which expects 100% reads of individual cases on a conveyer, today reads between 50% and 90% of tags going through on a full pallet.
Some of the early problems with tag reads were caused by factors such as tags that didn't transmit signals quickly enough, and too many tags being read at once, confusing readers about which signals to pick up. RFID implementers also still struggle with the fact that metal reflects radio frequencies and liquid absorbs them, making it difficult to read an RFID tag signal on containers with either material. That leads to what Miley Ainsworth, director of innovation and scanning technology at FedEx Services, terms the "place-very-carefully-and-ship" method, a time-consuming process of positioning the tag over an air pocket on the outside of a carton.
Most often, it's faulty or damaged antenna inlays inserted into RFID packaging labels that cause about a fifth of 500 labels on a roll of tags to fail. "Reading RFID labels shouldn't be about reliability," complains Kevin Ashton, VP of RFID tag reader manufacturer ThingMagic. Problems can occur at the reader level, too, when two or more readers are positioned close enough that the signal from one interferes with the signal from another.
While RFID chipmakers work on quality, many companies continue to look for the best combination of tags, label printers, and readers. Pacific Cycle has spent nearly $1 million on RFID since 2003, including three combinations of readers, tags, and printers. IS director Mathews expected perfect read rates by now, but the best it gets is 70% to 80% for cases. RFID tags that adhere to the new Gen 2 higher-frequency standard that sends data at up to 640 Kbps should help read rates, but he doesn't expect to see them in use for three to six months.
Ah, standards: Finally, there's some good progress to report. EPCglobal approved specifications for Generation 2 tags in December and just last month issued Gen 2 conformance certifications for some readers and chips from Alien Technology, Applied Wireless Devices, Impinj, Intermec Technologies, MaxID Group, Symbol Technologies, and ThingMagic. The group had its share of infighting over RFID-related standards, including which intellectual-property patents would be royalty-free.
Yet things are easing. Action groups within EPCglobal last year started developing standards to define reader management, a reader protocol, and tag-data translations. In 2006, EPCglobal will begin rolling out specs for more than 50 business processes to simplify transactions. Its first software standard, the Application Level Events, was released in September. It defines the process to collect, filter, and route raw data and translates it into a consistent format that multiple parties can share over the EPC Information Services network. The network will give more companies access to "clean, shareable data," promises Mike Meranda, EPCglobal's U.S. president.
Cost Factor
Now, if only unexpected deployment costs didn't continue to bedevil RFID adopters. Annin originally budgeted 20 cents for each RFID tag. VP Beard says he based that on a conversation with an RFID chip vendor, but the vendor was vague about what he would get for that price. He then learned the 20 cents didn't include the label, which adds another 20 or 25 cents to the price.
RFID doesn't work well in Gallo's distribution model, CIO Kushar says
But chip prices are falling, at least in large batches. Alien Technology last month unveiled a 44% price cut to 12.9 cents a tag for 1 million Class 1 EPC UHF tags. And UPM Rafsec says it's making available Gen 1 and Gen 2 inlays at less than 10 cents, for an order of just 50,000.
For FedEx, the costs and the error rates still are too high. "Why would I want to spend 28 cents per package to put on an RFID label when I'm already getting all the information I need to route and ship packages off the bar code?" Ainsworth says. RFID may have limited potential in the future, such as on returnable containers or shipments that require extra security, he adds.
Some companies face unique challenges. E.&J. Gallo Winery points to problems that occur in a three-tiered distribution model, which liquor vendors by law must use, VP and CIO Kent Kushar says. Gallo's distributors consolidate products from multiple vendors on one pallet, making it difficult to standardize tags and readers for all products they distribute. Gallo has joined with 24 companies, including Campbell Soup, Microsoft, and Wal-Mart, in creating the Information Technology Research Institute's RFID Research Center at the University of Arkansas to help fix RFID's problems.
There have been difficulties in the pharmaceuticals industry getting RFID tags small enough for pill bottles. It's "nearly impossible to print information on the tags because they're too small to go through an automated label applicator," says Randy Dunn, director of RFID at ADT Security Services Inc., the Tyco division that provides services to Mallinckrodt Pharmaceuticals, a Tyco health-care unit that ships RFID-tagged drugs to Wal-Mart.
Wal-Mart execs know all the problems its suppliers have had to tackle. And they remain confident that championing RFID was the right thing to do. "If we would have waited for all the numbers to prove RFID, we might not have taken a stand, and it might not have moved in 10 years," CIO Linda Dillman said at the InformationWeek Fall Conference in September.
Thirty years ago, many in retail and consumer goods considered bar codes costly, technically flawed, and unnecessary. It's a perspective RFID opponents point to often. "Today, bar codes are on 87% of the items in supermarkets, and the adoption rate of RFID has already proven to have caught on faster," says Peter Regen, VP of global visible commerce at Unisys Inc. Skeptics counter that it took 10 years for bar codes to be fully accepted, and they weren't trying to unseat an established technology.
Even for those who fully believe in RFID's potential to revolutionize supply chains, the intense interest can be wearing. With a smile, Wal-Mart's Dillman told InformationWeek conference attendees that she was "tired of talking about RFID." But considering the questions that still swirl around deployments, it's unlikely the conversation will be over any time soon.
|
