Teradata

By now, most major manufacturers have completed their first wave of RFID pilot projects and mandated compliance efforts; however, very few expect to recognize a positive return from their initial investments. Although these early investments are not entirely lost, many early adopters are wary of funding additional waves of RFID projects without a clear path to positive returns. The days of RFID as a research project are quickly fleeting. This paper highlights several opportunities for early adopters to unlock the hidden potential of RFID-enabled data and justify future waves of RFID investment.

Driven by Wal-Mart® and U.S. Department of Defense mandates, more than one hundred major manufacturers have completed their first wave of RFID UHF EPC pilot projects and compliance efforts. Yet, these early adopters are wary of funding additional waves of EPC investment without a clear path to positive returns. The days of Passive UHF RFID and EPC adoption within a compliance-only framework are quickly fleeting.

Increasingly obvious is that positive returns must be found by leveraging the inherent technological advantages of RFID over standard bar coding. Non-line of sight data capture and simultaneous tag readability represent two such advantages of RFID as read rates begin to approach that of bar codes. These technological benefits can result in higher productivity and throughput while offering opportunities for more timely and frequent data capture since human intervention may no longer be required to acquire data.

The ability to dynamically add, change, and remove information from RFID tags is another technological differentiator versus bar codes. Business benefit may be derived from this capability when valuable sensor (e.g., temperature or moisture) data are added to tags as a quality assurance measure. Information added to tags can also support business requirements for electronic pedigrees and maintenance records, or eliminate the need for a network connection in order to look up item information.

Finally, although bar code technology is capable of supporting the unique identification of items, it is the co-adoption of RFID UHF technology with the Electronic Product Code (EPC) which is giving companies the opportunity to perform serialized item tracking. While unique identifiers can greatly improve data quality and shipment accuracy by reducing duplicate and missed scans, companies will find even greater benefits through the analysis of this serialized (i.e., instance specific) item data.

As we consider RFID's technological benefits over bar codes, we can see the majority of the business benefits are likely to be derived from enhanced operational data generated by RFID technologies. And, as many in the industry have already discovered, the ability to transform these data into actionable information is the greatest challenge facing RFID adopters. Thus, it is the intent of this paper to highlight several opportunities for early adopters to unlock the hidden potential of RFID-enabled data and further validate continued waves of RFID investment.

Although RFID has been around for more than 50 years, the early adopters of Passive UHF RFID have often found themselves in uncharted territory. At most manufacturers, initial programs were deemed 'slap & ship' compliance efforts that focused on basic compliance at the lowest possible investment. These programs centered on the fundamentals of RFID technology at the tag, antenna, and reader levels within the boundaries of existing business processes and infrastructure. Some leading companies, however, saw their RFID programs as an opportunity to automate and re-engineer processes (especially within the warehouse) to leverage RFID's technological benefits as a way to improve productivity. While tangible benefits have been accrued from such productivity-centric implementations, the promise of a meaningful return on investment from RFID has gone mostly unfulfilled.

RFID adopters are rapidly realizing that the key to extracting real value from RFID is to harness the mass volumes of data it generates to improve visibility and business insights. This has been widely and effectively used in many other parts of the radio spectrum over many years and is widely acclaimed by those whose tracking, tracing, and asset visibility programs have been enhanced by the use of RFID. However, these applications have typically resided within closed-loop or one-to-one relationships, while EPC sponsored use of RFID is found in open, multiparty supply chains. If the promise of EPC RFID is to be realized, it must deliver the ability to track and trace items and assets throughout their life cycle, providing actionable information to decision makers/agents across company lines in time to prevent issues and take advantage of opportunities.

A first step towards harnessing RFID for improved visibility is the integration of RFID master and read data into a single data repository, sometimes referred to as the EPC-IS (EPC Information Services). If RFID read data are simply linked to local applications for operational support, the benefits of serialized data for track and trace functionality will be lost. To gain full benefit, these data must also be directed (through middleware) to a single database capable of integrating RFID reads across locations and event types to be associated with relevant master and transactional data.

For example, what good are data from an RFID read event if the product, location, and transaction information associated with that event are held in separate databases? While knowing that a particular group of cases moved through a dock door is good, knowing the shipment and order information associated with those cases must be better. This requires a company to take an enterprise approach (write once, use many) to RFID data management while eliminating the costs and errors associated with data replication.

This is not to say that RFID middleware should not route data directly to operational systems; instead, it implies that doing so can only improve an organization's reflexes and muscle memory. Just like living organisms, companies delegate predetermined sensory responses to operational extremities. For example, if you grab your morning coffee and it is too hot, your hand will pull away based on reflexes without your even thinking about it. However, it is the distinct advantage of intelligent organisms that sensory information is delivered to a central repository (a brain) where it may be committed to memory and analyzed in context through association with both prior sensory events and data from other sources. For example, if hot coffee spills in your lap as you are driving, your brain may determine a more appropriate split-second response than jumping out of your seat while speeding down the road.

Consider the consequences of storing data from each of our five human senses in separate brains. That's the corporate equivalent of storing RFID read data, order data, shipment data, invoice data, and inventory data in separate databases. And, where would the customer, product, and location master data be stored? How about retail point-of-sale data? This silo data model leads to memory gaps, response delays, and under-informed decision making.

In order to fully leverage RFID data, certain data governance rules must be applied to close the gaps created by imperfect read rates and bidirectional item flows. Although there are many types and combinations of inference logic that can, and should, be used, the correct approach depends heavily on the business scenarios they are intended to support.

Two basic examples are provided for your reference:

Association – If a pallet stacked with tagged cases is shipped and any acknowledgement of receipt is given, then all cases are assumed to have arrived at the destination. This logic may be applied even if the specific EPCs associated with the pallet and cases go entirely unread. All that is required is a simple confirmation of shipment receipt. Such associations may also be restricted to situations where a minimum threshold of cases must be read before the remainder is inferred to be present as well. For example, if at least 50% of the cases on a pallet are read, then assume any unread cases are also present.

Deduction – If ten cases of product are shipped, but only eight of those cases have their EPCs read upon receipt, then it may be wise not to immediately assume that all ten arrived. Instead, read events further downstream would be actively monitored for the missing EPCs. Should the EPCs reappear later, deduction logic could trigger a system update to acknowledge their initial receipt as part of the original ten-case shipment. This is clearly a stronger form of inference than association and, when both are in use, may be used to override associative logic.

No matter what type of logic is used to fill data gaps caused by less than perfect read rates, it is important to use that logic to update historical data in the system of record. It is also critical that updates made through inference receive special identifiers so that end users know that an inference (rather than a confirmed read) was made and which type of inference logic was used. Such information may be vital when disputes related to shrinkage, deductions, handling, and/or warranty occur.

The use of inference logic in support of track and trace capabilities is particularly important in industries with higher counterfeit and gray market activity. In the pharmaceutical industry, for example, each instance of a product may require an 'electronic pedigree' that contains all information related to that item's origin (e.g., batch) and life cycle (e.g., every handling event) for a ten-year period.

Although bar codes have only been around for about 30 years, they have become such an integral part of business transactions that they are often taken for granted. And so are their implications.

Because bar codes require proper orientation to be scanned, they typically require human interaction during the data capture process. And, since human capital is a scarce and costly resource, companies tend to limit data capture activities as much as possible — typically only when a transaction is taking place. But RFID can enable autonomous data capture to give visibility into an item's status at more frequent intervals. A perfect example of this is the way that Wal-Mart is using RFID to gain visibility into inventory movements from a store's backroom onto the retail floor.

This opens the door for a 'transactionless analytics'1 approach to preventing out-of-stocks by combining improved awareness of inventory on the store shelves with detailed demand forecasts, consumption histories, and upstream inventory. Analytics run against these data provide store personnel with prioritized backroom pick lists to replenish SKUs with the highest out of stock probability — and it is only the beginning. Another consequence of the traditional bar code paradigm is the assumption that differences among like items are insignificant. Of course, the makers of Tylenol probably felt the same way until the now infamous recalls of the 1980s. It is the memory of that incident that has driven Johnson & Johnson to take a lead position in the adoption of RFID and the embedded Electronic Product Code (EPC).Why? Because the EPC gives each instance of an item a unique identifier.

This serialized code can enable the near perfect traceability of items through an almost limitless number of batch splits, lot merges, and handling events. More important is the association of item-events with contextual information such as document numbers, material handlers, and adjacent items.

The bar code paradigm also ignores process inefficiencies and inconsistencies that can have a real impact on system performance. For example, most people assume that inventory flows through a warehouse or store in a FIFO (first-in, first-out) order, yet almost everyone has scavenged through packaged foods or milk at a grocery store in order to find a later expiration date. In addition, products are often misplaced in a store's backroom or left sitting on warehouse shelves that are difficult to access, but, without unique identifiers on those items, there is no way to know how frequently this occurs or what the business impacts may be.

Serialized item data are an invaluable aid to the measurement of cycle-times, which is a prerequisite to improving them. These data can also help uncover process bottlenecks and breakthrough opportunities typically disguised by the un-serialized data provided by standard bar codes.

When FIFO is assumed, one may conclude that cycle-times vary across a rather normal distribution. However, the calculation of actual cycle-times with data coming from the movement of serialized items usually reveals a very different story.

Armed with a detailed understanding of cycle-times and their variability, companies are empowered to extract excess safety stocks from the supply network while simultaneously reducing out of stocks.

Serialized item data can also reveal inefficiencies and opportunities related to the routing of items through a supply chain. Almost every item can take multiple routes from a given origin to a final destination, and with each route, significant differences in logistics costs, product quality, and delivery performance may appear. The only way to fully monitor and control these flows is with serialized item data tied back to each handling event and transaction.

Finally, the opportunity to dynamically add and modify data stored within an RFID tag is another reason to break from the bar code paradigm. Data from sensors (e.g., temperature or vibration) may be written to RFID tags at regular intervals or when certain thresholds are exceeded. These data may be used to dynamically adjust product expiration dates (e.g., fish filets), or to predict the likelihood of product failure before installation (e.g., semiconductors). Storing additional data on RFID tags can also alleviate network bandwidth constraints and improve the productivity of mobile workers using handheld devices. At the end of the day, however, a company's ability to put these additional data in analytic context will determine whether or not the benefits received are simple efficiencies or procedural breakthroughs.

Fully implementing RFID can be a significant investment — especially if the implementation is confined by existing infrastructure and processes. Non-line of site data capture and simultaneous tag readability provide a number of opportunities to change facility layouts, material flows, and the business processes that govern them. Yet a greater opportunity may exist by leveraging serialized data to break down the information silos that prevent cross-organizational collaboration and optimization at the system level.

One of the barriers to system-level optimization is organizational-centric performance measurement. When each function independently optimizes its own performance, system performance as a whole suffers. Organizations will always attempt to manage what they control or have responsibility for, but so many of today's business problems occur in the gaps that exist between each organization's span of responsibility. Without serialized data to weave together specific records of causes and effects across organizations and processes, measuring, and therefore improving, the gaps can be an impossible task.

Just imagine trying to calculate the actual profitability of each individual unit (or case) by taking into account detailed measures of direct and indirect costs associated with materials, production, logistics, planning, marketing, sales, and service. (Most companies cannot even fathom what it would take to truly measure invoice level profitability.) But, there is value in doing so. And, it is critical to driving fact-based cost and service tradeoffs both within and across organizations.

The Electronic Product Code is more than a number; it is a powerful change agent capable of stringing together disparate organizations and disjointed activities. Companies that take a more functional approach to RFID implementations will often be disappointed, but those companies that see and leverage RFID adoption as an opportunity to drive system-level changes will unlock the ROI in RFID.

The technological advantages of RFID enable its adopters to better relate real-world events to the virtual-world of information technology and business intelligence. Simply put, RFID is an enabling technology that provides businesses with a new, richer source of data about their operations. While these data can improve efficiencies within existing operations, greater returns can be found when RFID is used to drive system-level changes.

To achieve these system-level benefits, companies must move beyond operational implementations of RFID that rely on middleware as the key enabler to improved execution. A few industry leaders are already realizing the benefits of integrating RFID with information from a number of transactional and master data sources to create a contextually rich analytic environment.

This approach not only helps to reduce RFID expenditures by better quantifying the benefits of specific RFID investments (e.g., tagging of particular products or asset types, modification of select processes, throughput capacity, and labor-hour improvements), but it also creates a powerful platform for the:

• Creation of high-context dependent (non-reflexive) alerting and event management
• Use of serialized data in the re-engineering of cross-functional business processes
• Provision of information to trading partners for fact-based collaboration and decision making

While the applications and benefits of RFID will vary greatly from company to company, these companies can rely on one simple truth: The key to value from RFID is what you do with the data. By putting RFID data in context; using inference logic to fill the gaps; challenging the bar code paradigm; and leveraging RFID as an opportunity to drive system-level change, your company will be well on its way to generating positive returns from its RFID investments.

To find out more about how a Teradata RFID solution can help you maximize your decision-making capabilities and grow a stronger, more productive business, contact your local Teradata representative.

Jared Schrieber brings a wealth of expertise to his lead role for supply chain management and RFID solutions in the manufacturing industry. He holds a master's degree in Logistics (MLOG) from Massachusetts Institute of Technology (MIT) and a bachelor's degree in Supply Chain Management from Arizona State University. While at MIT, Jared was engaged in advanced and applied research on RFID. He currently consults numerous companies about how to harness the potential of RFID data to transform their operations. His professional experience includes retail management, supply chain consulting, and logistics management.

1 Transactionless Analytics: Most analytics today are run on transactional data, but RFID is opening an entirely new realm of analytic opportunities by providing data about items and events in-between business transactions.