Many companies today are aggressively employing analytics—the systematic use of quantitative and statistical decision methods—in their businesses. There are many different application domains for analytics, ranging from marketing to human resources to finance. It is only natural, then, that the next generation of supply chains should incorporate a higher and more sophisticated level of analytics.
Applying analytics in supply chain management is not a new idea. The U.S. military adopted a variety of logistical models in World War II, and companies adopted related approaches in the postwar period. UPS, for example, established a logistical analytics group in 1954. Since then, many companies have successfully employed analytical approaches to distribution networks, inventory optimization, forecasting, demand planning, risk management, and other applications. Large retailers, such as Wal-Mart Stores and Target, have had considerable success with supply chain analytics, often working in collaboration with suppliers. And carriers like UPS, FedEx, and Schneider National wouldn't dream of managing their operations without a variety of analytical models.
Yet supply chain-related analytics activities have plateaued in many organizations in recent years. Other than the occasional re-tuning of supply networks that has principally focused on cost management, companies have not taken advantage of all that supply chain analytics can offer to their businesses. Further, even when analytical tools are available to front-line supply chain personnel, the tools often go unused because of a lack of skills or understanding.
We believe that there will be a set of new frontiers in supply chain analytics that will lead to dramatically higher levels of performance. If companies are to achieve these rewards, however, they will have to be more ambitious in their analytical goals and investments. In this article we describe a number of relatively new domains for supply chain analytics as well as the opportunities and primary obstacles for each. We also describe several ways in which the day-to-day usage of supply chain analytics will change in the future.
Connect demand and supply in real time
One of the most important attributes of next-generation supply chain analytics is that they will address issues beyond the supply chain. To optimize operations, companies need to link their supply chains with metrics and analytics on the demand side. For example, at the simplest level, price changes or promotions for products will change demand and hence the required supply of those products. Similarly, changes in the availability of products and components should be reflected in marketing and sales processes.
This integration of supply and demand was pioneered in the 1990s by Dell Computer, which was able to suggest to call-center customers ways to shorten delivery time or take advantage of excess inventory. This was mostly dependent on human decision making: manufacturing supervisors would track supply levels and notify sales and marketing managers, who would then promote or downplay particular items and configurations based on their availability. But in a real-time, online business environment, companies will need to have analytical models in place that will continuously integrate supply and demand without human intervention. Such models would, for example, automatically extend offers and promotions to customers based on the availability of inventory and components. There has been a shortage of initiatives in this area since Dell's pioneering work, but the direction for future innovations is clear.
The analytics needed for such models are not terribly difficult, though they would require considerable iteration and tuning. The primary obstacle to implementation generally is a lack of collaboration among multiple transaction systems, in a way that allows companies to make informed decisions in real time.
Analyze supplier risk
Many companies recognize that the success of their operations is highly dependent upon their suppliers. Yet supplier risk analytics have hardly moved beyond simple metrics and reports in most organizations. The most sophisticated approaches to supplier risk monitoring and management—used by companies that heavily depend on external suppliers and contract manufacturers, such as Cisco Systems—are only somewhat more analytical.
One example is the creation of a supplier resiliency score based on several variables. The variables are based on logic (for instance, reports of bad weather near suppliers' manufacturing locations). If the variables or the overall resiliency scores suggest a problem, companies can then pursue secondary sourcing or work with existing suppliers to identify alternate locations. These scoring models increasingly incorporate relatively subjective factors, such as perceived economic and political risk. But while supplier risk and resiliency scores are undeniably useful tools, with few exceptions they are not yet based on statistical analysis.
Of particular interest to many companies now is whether critical suppliers that weathered the last economic downturn will be capable of meeting increased demand during an upturn. Analytic tools that incorporate public, third-party data can help companies assess this risk.
As companies accumulate more experience with supplier risk, they can begin to create predictive statistical models that are based on actual supplier failures. This would, of course, require tracking and analyzing a sufficient number of actual supplier failures to allow them to accurately identify attributes associated with failure.
Interestingly, the current leaders in statistically assessing supplier risk generally are not the manufacturers but the firms that insure them against such risk. Because the insurance industry has a strong actuarial tradition, firms such as Aon and Marsh have developed statistical models of the likelihood of supply and supplier risks. The key variables considered in these models are the frequency and severity of those risks.
Take advantage of sensors
One of the primary drivers of analytics in organizations is the availability of extensive data. As their use expands, new sensors—in particular, radio frequency identification (RFID)—will make dramatic amounts of data increasingly available for the next generation of supply chains.
For more than a decade, supply chain managers have been bombarded with warnings that RFID devices and networks will change their lives. Thus far, however, the high price of RFID technology has prevented widespread deployment from taking place. But prices for RFID tags and readers continue to fall, albeit slowly, and the adoption rate is gradually rising. At some point in the next several years, most manufacturers and retailers are expected to deploy some degree of RFID capability. When that happens, a great deal of RFID-generated data will be available for analysis. Initial applications using RFID data will primarily be transactional, but shortly thereafter organizations will want to monitor and optimize the efficiency and effectiveness of their RFID networks. This set of applications will demand the use of sophisticated supply chain analytics.
Some companies have employed RFID analytics for several years. For example, Daisy Brand, a dairy products manufacturer in the United States, began using RFID analytics in 2007 to track how long it takes products to reach the store shelf as well as replenishment rates. Prediction of replenishment rates is particularly important during promotions. In addition to RFID data, Daisy Brand also makes extensive use of Wal- Mart Stores' Retail Link data, which provides suppliers with weekly point-of-sale and inventory information, in its analyses.1
Sensors for more expensive and substantial supply chain assets are already in wide use. Some major carriers, for example, are deploying geographic positioning system (GPS)-based telematics devices in trucks and trains. These devices provide a wide variety of data about driving behavior, speeds under various conditions, traffic, and fuel consumption. Companies such as UPS and Schneider have already employed telematics data to redesign logistical networks in whole or in part. UPS, in fact, is using telematics data to redesign and optimize its entire delivery network for only the third time in its more than 100-year history.
Other types of sensors are likely to lead to a flood of additional data—and opportunities to analyze it. RFID and telematics sensors primarily track location, but so-called ILC (identification, location, condition) sensors can also monitor the condition of goods in the supply chain. ILC sensors monitor such variables as light, temperature, tilt angle, gravitational forces, and whether a package has been opened. They can transfer data in real time via cellular networks. Obviously, the potential to identify supply chain problems in real time and take immediate corrective action is greatly enhanced with this technology. We have only begun to consider how analytics might be used to enhance the value of ILC-derived data.
Improving analytical "literacy"
The next-generation approaches to supply chain analytics involve not only new applications but also new ways to ensure that analytics are used to make strategic and tactical decisions. Unfortunately, better decision making in supply chain management is often hindered by the inability of managers and front-line personnel to understand and apply analytical models.
We have encountered several companies that had considerably upgraded the analytical capabilities of their information systems (for example, by adding advanced planning and optimization modules for enterprise resource planning [ERP] systems) but had made no changes in associated personnel or their analytical skills. As one supply chain manager told us, "We need only half the people to do the work with these new tools, but they need to be twice as smart." For supply chain personnel to become smarter about analytics, they must be educated about analytics and their implications, retrained, or in some situations even replaced.
There are a variety of approaches to achieving the desired level of analytical literacy. The motor carrier Schneider National, for example, has developed a simulation- based game to communicate the importance of analytical thinking in dispatching trucks and trailers. The goal of the game is to minimize variable costs for a given amount of revenue while maximizing the driver's time on the road. Decisions to accept loads or move empty trucks are made by the players, who are aided by decision-support tools. Schneider uses the game to help its own personnel understand the value of analytical decision aids, to communicate the dynamics of the business, and to change the mindset of employees from "order takers" to "profit makers." Some Schneider customers have also played the game.
Another way to facilitate the understanding of supply chain analytics is through simpler applications with narrow functionality. Increasingly referred to as "analytical apps," these tools are similar to the applications found on smartphones. They support a single decision and often are industry-specific. Several business intelligence and analytics software vendors are introducing them, and they promise to make the use of analytics much simpler and available to users who do not have extensive analytical or technological skills. Analytical apps that have already been developed for supply chain functions include tools for supplier evaluation, inventory performance analysis, transportation analytics, and transportation contract compliance. There undoubtedly will be many others over the next several years.
Perhaps the only way to guarantee the use of analytics in supply chain management is to embed them into supply chain-oriented systems and processes. No human would be involved in the decision unless there is an exception. For example, certain supply chain decisions made at least partially on the basis of statistics and probability (such as available-to-promise inventory, or the likelihood that an ordered product will be returned by the customer) could be embedded in an order management system. Vendors of ERP systems expect to have such capabilities in the next several years.
The future of supply chain analytics
The use of such tools as ERP systems, the Internet, RFID, and telematics is becoming more common, and more organizations are generating considerable amounts of high-quality data. Now that companies have more and better data than ever before, it is only natural that they would begin to use it to analyze, optimize, and make predictions about their supply chains.
The most common analytical activities thus far have been descriptive—straightforward reports about what has happened in the past. But in future supply chains, we expect to see more prediction and even prescription—that is, optimization and testing models that tell supply chain managers what they should do to improve performance.
Employing emerging supply chain technologies and process improvements has always been an important path to competitive advantage. We believe the next major approach to supply chain-based competition will involve the extensive use of analytics.