Unleashing the Potential of Supply Chain Analytics

Hanesbrands Inc., a manufacturer and marketer of basic apparel based in Winston-Salem, North Carolina, is using analytics to close the gap between insight and supply chain responsiveness. For example, the company recognizes that knowledge of a shortage of men’s T-shirts two weeks from now is of no benefit if the minimum lead time necessary to acquire more T-shirts is four weeks. So Hanesbrands is turning to machine learning to design predictive models to sense supply chain issues in time to execute prescriptive measures.

The predictive models incorporate supply chain data from external and internal sources to determine the likelihood of an inability to satisfy demand at a particular time. Once an impending supply chain issue is detected, a prescriptive action can be launched to mitigate it. For example, if an inventory outage is predicted, Hanesbrands will assess available options, depending on the time available for a response. The actions prescribed could involve a change in mode of transportation or a resequencing of manufacturing and purchase orders. Concurrently, the prescriptive model works to balance mitigation cost with the benefit to the company.

If that’s not how supply chain analytics works in your company, you’re not alone. Only a handful of companies have developed an ability to react quickly to supply chain signals. Despite analytical advances, many companies still do not fully understand the sources of their competitive advantage, even though they now have more data than ever about their operations.

The results of a recent study of supply chain professionals we undertook suggest that despite a growing interest in using analytics to better understand complex performance relationships, many companies still struggle to develop this capability as a competitive resource. (See “About the Study.”) Although the supply chain is recognized as a particularly rich area in which analytics could be used to improve performance, supply chain analytics is still in the early stages of development and implementation in many companies.

The first stage of our study, a large-scale survey of North American businesses, found that the primary users of supply chain analytics are operations (49%), marketing/sales (25%), and supply chain management (24%). The users of the supply chain analytics efforts rely on outputs that predominately involve only basic statistical analyses (44%) and forecasts (43%), indicating that few companies have sufficient skills on tap to use more sophisticated tools. In addition, we found that several common obstacles prevent companies from developing and implementing supply chain analytics capabilities: inaccurate data (42%), the cost of implementation (31%), unavailability of data (31%), and a lack of timely data (30%).

The effect of unreliable data on supply chain analytics capability cannot be overstated, because it affects the data that is shared with key customers and suppliers. The majority of survey respondents have attempted to use supply chain analytics to provide actionable insights or prescriptive recommendations to improve operational performance in functional activities (such as inventory levels, supplier lead times, or forecast accuracy). However, the time currently needed to respond to these insights is too long to be effective. On average, most companies require several days to a week to change their manufacturing batch size for a particular product or to increase the purchase order size from a critical supplier. Less than 25% of companies surveyed can respond immediately to insights from supply chain analytics.

Currently, supply chain analytics are used more frequently to reduce costs (56% of respondents) than to increase revenue (38%). Analytics tools are often used to forecast supplier lead times. They can also be used to identify the best location from which to fill an order today so as to improve customer service levels while maximizing the likelihood of filling orders tomorrow. Among the survey respondents, customer service is the most popular application area (61%) for analytics tools. Two areas where analytics tools are currently utilized least are stock outs (35%) and fill rates (40%); this is somewhat surprising given how integral those areas are in managing customer service in the supply chain.

Response Speed Is Key

We believe that in the future, companies will need to not only advance their supply chain capabilities but also need to be able to act on the recommendations of their analytics engine in a timely way. In our view, this speed of response, which we call “analytics insight cycle time,” is likely to become a key success factor in the next few years.

Our research identified five levers critical to achieving a highly efficient level of analytics insight cycle time.

1. Supply chain analytics initiatives need a top-down mandate. The analytics culture of the organization sets the stage for analytical success. Executives in the C-suite must promote and support the supply chain analytics initiative in both word and deed to create an atmosphere where analytics becomes an integral part of the fabric of the organization. “A top-down mandate is key,” according to Baptiste Lebreton, operations research manager in the data science group at Eastman Chemical Co., a materials and specialty additive company based in Kingsport, Tennessee. During our interviews, we heard that without a top-down mandate from senior executives, it is very difficult for companies to advance to a higher level of supply chain analytics capabilities. David Dittmann, director of business intelligence and analytics services at Procter & Gamble Co., stressed that “it is impossible to win over thousands of people one small analytics victory at a time. Analytics must be a top-down mandate to succeed from an organizational perspective.”

2. The simpler the model, the more likely it is to be used. To minimize analytics cycle time, model simplicity and understanding are important. In fact, Gary Kinney, associate director, global supply chain analytics at Procter & Gamble, noted that “if a user does not understand an analytics model, they will not use it. Moreover, models viewed as unnecessarily complex enigmatic ‘black boxes’ are not successful. For example, if an analytics recommendation is perceived as disruptive, it contradicts managerial instinct. If users do not understand a model’s seemingly contradictory results, they will not use it.” Richard Daigle, group director, automation and analytics, Coca-Cola Refreshments USA Inc., added that “analytics supply chain models must be simple and transparent, yet effective; to gain necessary buy-in, analytics professionals must be able to explain the model in terms that anyone in the organization can understand.” In other words, the simpler the model, the greater the likelihood that people will use it.

3. Domain and business knowledge are both essential. Without a fundamental understanding of the business in which a supply chain analytics model will be deployed, developing the right model is a tall order. The belief that “inventory is inventory,” and it doesn’t matter whether the business is selling lawn mower parts or making chocolate chip cookies, is simply not true. Executives told us that a complete and thorough end-to-end understanding of the key business processes and their effect on the bottom line is critical to supply chain analytics success. Without it, one “can end up with nothing more than fun and interesting facts,” observed Ben Martin, chief officer, advanced analytics and global planning at Hanesbrands.

Gulfstream Aerospace Corp., an aircraft manufacturer based in Savannah, Georgia, and a subsidiary of General Dynamics Corp., realized that without sufficient domain knowledge, an external vendor would face an insurmountable challenge when tasked with creating an inventory management system for the company. Thus, the decision was made to develop its own system in-house. Likewise, Dittmann and Kinney at Procter & Gamble acknowledged that it is “critical to lead analytics in-house, not delegating to a consulting firm, because consultants don’t have as comprehensive an understanding of the business as the people in the business itself.” In fact, they stated, “If the SCA [supply chain analytics] professionals are not properly wired into the business, more problems than solutions will result.”

4. The organization must trust the numbers. To successfully operationalize supply chain insight in a timely fashion, users must trust the insights that supply chain analytics generates. Trust can be built through a closed-loop change management effort that is centered on performance metrics that accurately reflect the current state of the supply chain system. Once such clear metrics are in place, new supply chain analytics systems can be launched and their performance relative to the old system gauged immediately. As users compare metrics, their trust in the analytics increases. The launch of Coca-Cola Refreshments’ supply chain analytics model to increase on-shelf availability of its product, for instance, was a big success due in large part to the deployment approach centered on a change management process that cultivated trust in the analytics. Any time users did not “trust” the system and therefore chose to override the analytics model, they were provided feedback on whether their manual overrides led to an improvement or a decline in on-shelf availability, thus creating a truly closed-loop feedback system that taught users to trust the analytics. A formal change management process significantly improves the likelihood of supply chain analytics success not only at Coca-Cola but across the board.

5. Companies need mechanisms that help their analytics professionals develop domain knowledge. The availability of trained analytical talent affects the supply chain analytics cycle time. For example, Eastman Chemical embeds new analytics hires in operations before moving them into a corporate supply chain analytics group. Gaining experience across business units allows the analytics hires to cultivate deep fundamental domain knowledge that is then integrated with their expertise. Lebreton stated, “At Eastman, the analytics professionals must possess sufficient business knowledge before stepping into the supply chain analytics role [or] else a lot of time is wasted.”

Hanesbrands takes a different approach to diffusing and integrating supply chain analytics talent. There, the analytics group operates as a center of excellence that leads analytics efforts with business functions such as supply chain, distribution, marketing, and sales. The group began as a small collection of analytics professionals from the consulting industry with strong analytics skills and six to 10 years of experience. As the group has expanded, new hires now acquire needed business knowledge through two mechanisms: (1) a strong mentoring program in which experienced analytics professionals guide new hires on project teams diffused throughout the company, and (2) a unique problem-solving framework designed specifically for Hanesbrands. New team members utilize the framework while serving on analytics projects alongside their mentors. Martin pointed out that “this process helps new hires learn the business and how to attack a problem.” As the new hires’ understanding of the business grows, it is integrated seamlessly with their analytics skill set.

Next Steps

Are forecasters wrong to believe that analytics will soon become an essential basis of competition and growth in the future, enhancing productivity, reducing waste, and increasing the quality of products and services? No.

In our research, we found that, in general, company performance improves as supply chain analytics grows more sophisticated. However, for supply chain analytics to become an important competitive resource, companies must reduce the time it takes to identify a supply chain problem or opportunity, perform the appropriate analysis, and transform the insights into action. This can be achieved only through a shorter insight cycle time that encompasses every aspect of supply chain management. Only then will supply chain analytics become a true tool for establishing a long-term competitive advantage.