Adding intelligence to an RFID pilot

META Trend: During 2004/05, external pressures to leverage new technologies (e.g., RFID, UCCnet), provide customized services, and improve visibility will drive organizations to upgrade supply chain execution applications (e.g., warehousing, transportation, manufacturing). Concurrently, international trade, security, and compliance pressures will motivate organizations to upgrade global trade, health/safety, and contingency planning solutions. Through 2008, organizations will merge information processes among CRM, SCM, and PLM applications to holistically scrutinize demand/revenue flows across customer and product life cycles.

Many current RFID pilot projects focus on achieving 100% read rates at speeds set by bar-code technology. The center of attraction for such pilot projects is the proper tag placement on cases and pallets, as well as the proper configuration of the pallet to enable the achievement of perfect read rates.

Integrating RFID read capabilities to enterprise applications is yet another component of a successful RFID pilot test. However, our research indicates that many of the RFID pilots underway do not include the impact to the data warehouse and the business intelligence applications as part of the pilot test. By ignoring these, the pilot test’s accuracy and business impact assessment can be impaired, significantly affecting the cognitive capacities of an adaptive organization and undermining the long-term success of the enterprise’s RFID deployment.

This oversight is a result of focusing on getting the tag and readers to work properly and having overlooked the fact that it is the decisions based on the information generated (bar code- or RFID-based) that will improve overall business operations. Alternatively, it may be due to concerns about pilot project costs. Indeed, the industry has been “wowed” by estimates of more than seven terabytes a day of Wal-Mart item-level data, and this may explain why the data warehousing teams stand by waiting for data from the pilots, instead of such teams being actively involved in the design of the pilot itself.

We believe it is time to take a closer look at estimating the RFID impact to a data warehouse. To proceed, we must recognize that 2005 will not be the year of item-level data and that item level RFID will be two to three years out (2007+). Furthermore, we must recognize that case/pallet-level data will assist us with determining the amount of data that is practical to capture and to use.

As with all technology, there is a point of diminishing returns, and this same axiom applies to RFID data. As with past data warehouse development efforts, there is discussion around the amount of atomic-level data to store in the data warehouse and for how long the data should be retained. Therefore, let us first review the amount of bar code-based data that is currently captured to determine how RFID may or may not influence the volume of data captured and retained.

For example, 20 cases of item X identified at a case level by bar codes will still be 20 cases with RFID. The only difference is that those 20 cases are uniquely identified at the case level with a product-level identifier that is accompanied by a serial number to identify which one of the 20 cases a single case is; we refer to this concept as serialization. Therefore, in this example, 20 cases (or pallets) with bar codes will still be 20 with RFID tags, and there is no change in the number of items added to the data warehouse.

The next area of estimating RFID data is the number of captures made while product moves through the enterprise. If an enterprise simply replaces its bar code readers with RFID readers at the same capture points, there will still be no increase in data for case and pallet movement, unless more captures are made due to increases in product throughput volume (e.g., 10 cases/minute becomes 20 cases/minute).

However, the real increase in data volume is the addition of more capture points (e.g., shelf readers) and an organization’s desire for improvements in product availability, visibility, and market-reaction capabilities.

To illustrate this point, one simple capture point that is added to an RFID project at a store level and at a distribution center level is the placement of a RFID gate for capturing trash that can contain empty or broken case/pallet tags. This addition of the trash gate is done to improve visibility into shrinkage numbers (e.g., damaged cases) and to record the disposal of any empty cases or pallets. This tracking is enabled by capturing the inbound receipt of a case/pallet and the outbound shipment of the same case/pallet in a distribution center (DC). If the case/pallet goes into the trash when it is supposed to go onto another truck, the RFID trash gate will capture that case/pallet as disposed or damaged and deduct it from the shipped quantity. Therefore, if five cases/pallets are inbound to a DC, five cases/pallets will be accounted as outbound shipped or some mixture of shipped, shelved, and disposed or damaged. Correspondingly, it is the number of additional captures that will increase the space requirements for a data warehouse, not RFID itself.

Indeed, the analysis of the capture points should be part of an RFID pilot, and the data warehouse team should contribute to this process. Data warehouse team members working with the applications team should determine the proper capture points required for product movement and analysis as part of the pilot. Moreover, this combined team will also determine if the data captured at these points is needed for an extremely short time, or whether it should be retained by the data warehouse because that data is valuable to the analytical capabilities of the enterprise. For example, a read to determine routing the product to its next step in the journey through the supply chain (e.g., place case/pallet X1234 on truck Y1234) may not be retained, while a read done to provide real-time inventory counts (e.g., 20 cases) for the next forecast cycle may be retained for trend analysis.

Bottom Line: RFID pilots must include an impact assessment of the data warehouse and should also include data warehouse team members. Without these key components, such pilots can severely impact the long-term success of the enterprise’s RFID initiatives.

Business Impact: Successful enterprises will incorporate proper analytical capabilities with improved RFID technology sensing capabilities to enable prescriptive situational analysis.

Related Download
Virtualization: For Victory Over IT Complexity Sponsor: HPE
Virtualization: For Victory Over IT Complexity
Download this white paper to learn how to effectively deploy virtualization and create your own high-performance infrastructures
Register Now