Felipe Molino, Director of Supply Chain Solutions Engineering,NFI
As leaders, we are met with very high expectations of our stakeholders. As much as we would like to be able to do everything, it is unusual to have the bandwidth required to work on all of our initiatives simultaneously. Therefore, we rely on our teams and trust them to conduct the analyses needed to support our decisions.
"The most successful supply chain analytics teams are made up of individuals with varied years of experience in different business areas who understand that modeling does not equate to executing"
Working at a 3PL has given me a new perspective on the types of skills needed to manage a supply chain analytics team effectively. NFI, for example, has long-standing relationships that have enabled us to mature not only as an operator but also as a strategic analytics partner. Our customers expect us to transcend algorithms and provide an executable business plan based on our operational expertise. Therefore, creating a solution is not as simple as “making a model.”
From my experience and conversations with other industry leaders, individuals can make significant contributions in many different ways. However, there are four categories I find many advanced supply chain teams have in common. This is not meant to be a comprehensive skills list, but rather a set of broad categories to help us evaluate internal expertise and identify potential areas for improvement.
Data Management and Database Skills
We have an unfathomable amount of data available these days. For instance, a typical SKU-level transaction dataset for one year’s worth of data will likely contain millions of records, which is well beyond Microsoft Excel’s processing capabilities. In addition, a single dataset may not contain sufficient information to complete a meaningful analysis.
Consequently, analysts and data scientists are spending the majority of their time finding, cleaning, and reorganizing data rather than analyzing it. In a time-sensitive environment, it becomes a challenge to finish a comprehensive analysis, and sometimes, businesses are forced to rely on high-level metrics, averages, and “good-enough” solutions.
Fortunately, there are multiple open-source and commercial data management technologies that allow us to process large and complex datasets. Advanced supply chain analytics teams can ultimately leverage these technologies to automate data workflows that enable them to spend more time analyzing rather than working with data.
Advanced Analytical Expertise
Companies have several business processes, such as demand planning, production, inventory allocation, and distribution that are connected through day to day business activity. Although companies are always looking to optimize these, there is an inherent limit to the overall benefit when optimizing a single business process. This is because “optimization” implies the best solution within a system, given a set of assumptions and constraints. Therefore, a system as interconnected as a supply chain must consider all business processes as a whole to achieve true optimization. The most successful supply chain analytics teams understand this interconnectivity and leverage optimization to design a solution that meets the needs of the end-to-end supply chain.
Other methods of advanced analytics include simulation and machine learning. Simulation involves studying how events, such as transactions, pick-ups, or deliveries, are influenced by time, costs, or other factors to find bottlenecks and other opportunities. Machine learning, on the other hand, enables companies to predict events that can happen within a supply chain to make changes proactively. The key to successful analysis is understanding that there are methods beyond what is typically done in Microsoft Excel today. It is also essential to recognize which of these methods is most relevant to the problem at hand and learning how to combine them to achieve the best solution. For instance, a team can connect a machine learning-enabled demand planning process with an SKU-level simulation model to predict what events will cause a stockout and when it will occur.
Although a solution may be “optimal” and show tremendous savings, it does not mean it is practical. As an example, your team created a model that shows 15 percent savings for closing a warehouse. The analysis may not consider that we have ten years left on the lease and a strong labor union in that building. This does not mean that the solution may not be implemented, but these are real operational issues that need to be accounted for, as they may offset the savings potential.
We need our teams to convert that virtual solution into something actionable. The most successful supply chain analytics teams are made up of individuals with varied years of experience in different business areas who understand that modeling does not equate to executing.
Communication and Relationship Building
While your team may have found the perfect balance between optimal and executable, you will still need to be able to sell your solution. In most businesses, there are multiple levels of internal and external stakeholders you need to go through before implementation takes place. In a situation where the analysis shows conflicting incentives among those involved, presenting data and models can be more of an art than a science. We need to remember that we are dealing with people; an executive will be looking for a much different set of criteria than a manager, or even another analyst. We need people that can speak to and present this information to individuals with varying backgrounds at each of those different levels.
A supply chain is a cross-functional discipline that requires a team to have both technical and soft skills. Few employees are experts in all of these categories, and they are typically your superstars. Recognizing the strengths of each team member provides a path to assembling the most effective project teams and conducting meaningful and executable analyses.
Does your team have all of these skill categories? What are some areas your team needs to improve upon? What are some specific technologies your team is using for modeling and data management? Have you identified someone in your team with all of these skills or the ability to learn them?