Expeditors International of Washington, Inc. (EXPD) Earnings Call Transcript & Summary

Similar, if you will, and just as you have probably experienced over the last couple of years, if you're coming on to the webinar, you are automatically muted and your camera is turned off. [Operator Instructions] And then also, we are going to answer the questions as we can. We may -- you may not see us answer it right away live, but we will address it when we get to the Q&A section, which will be at the end. And if we don't, obviously, we'll send a follow-up on that specific for you. At the end of the seminar, we will send a survey that we ask you to fill out. When you do, we have some resources that the Supply Chain Solutions team has put together, and they want to share those with you. So we thank you for those as well. Go ahead and as we get started, this particular webinar was put together by our Supply Chain Solutions team, and I'm going to introduce them in just a second. It's really going to be a great hot topic, I believe. So I believe you'll all enjoy it. I wanted to thank my co-host that I have on the screen as well. They're not the ones that are the speakers, but the ones you'll see on the camera. They're still on, and you'll see [indiscernible], who is our District Manager -- excuse me, District Sales Manager for Miami, and he is my counterparts and helps support in the Southeast. Raven Campbell is our Sales Operations Coordinator, and she is actually going to be your new face for the Southeast region for any of our seminars that are coming up in the future. You may see her name in your inbox quite often, whether it be a target marketing where we're trying to just make sure that you're getting the most recent information or if we have a request that we would like to invite you to attend, you'll see her name with that. And then we have Samantha Hurst, who is also helping support and she's from the Mid-Atlantic district, and we thank you for that team members, for those teams to come in. I'd like to also introduce our speakers for today. We have Melissa King, who is the Director of Supply Chain Solutions for the Americas. Melissa has been with supply chain solutions for 12 years, and she is over the team that you're going to be introduced to today as well. Shane Wood is our Regional Manager for Supply Chain Solutions, and he helps lead client engagement. He's also based in our Colorado office out of the Denver area. Melissa, I forgot to mention, is in Charleston, right where I sit. So we are happy to have Shane and Melissa. Shane has been with us for 17 years, and he's actually worked in supply chain solutions for the past 7. Hack Russell is our Project Manager for Supply Chain Solutions. He's been with Expeditors for 12 years, and he has experience in order management and our customs brokerage products and is currently a Project Manager for the Supply Chain Solutions Group. And for the sake of our webinar, I'm actually going to turn it over to Hack to start us off. Thank you.

Great. [ Rona ], thank you for that introduction. So supply chain design, it's how we procure and move goods throughout our supply chain in a thoughtful way to make sure that we can reach the performance measures that we set forth. And as this roles in our industry, just continue to get more and more specialized. We see our complexity increase kind of in parallel to that. So choosing the right mix of distribution facilities, transportation modes and service providers is a challenging task, but it has a profound impact on your bottom line. And then luckily, as these complexities increase, we are seeing more and more technology solutions to help support how we manage our supply chain. With that, we're going to go to our first poll. You'll see it in front of you. How recently has your team conducted supply chain design study? 1 to 3 years, 3 to 5 years, 5 to 10 years or what is the supply chain set? All right. As we wait for you guys to put in your answers, I'll continue. So for many years, logistics was just an afterthought, is often described as a necessary evil of doing business. And as such, supply chains existed as several independent silo functions. Goods move linearly through the supply chain from step to step, ultimately arriving at the final customer with that performances and costs of each of these nodes were reviewed separate from one another as well. In a lot of cases, the management took place with different teams, different companies. You have different compartments -- departments competing to meet their goals without the ability to affect decisions made in other areas. The first thing that I remember learning about supply chain, this came from a college professor of mine, was that logistics is the only sector of business where $1 saved means $1 earned -- $1 more to your bottom line. And what this means is that all other areas of your company, if you reduce your costs by $1, there's some sort of opportunity [ lost ] as a result of that savings. So in marketing or sales, if you reduce your expenditure by $1, you're going to reach a smaller audience, and with that, there's going to be a loss of sales. Or in finance, if you reduce your finance budget by $1, you're going to underutilize your capital deployment by some amount, meaning there's going to be a loss of return. But in supply chain, if you deliver your product to your customer at a lower cost, then your margin is going to increase accordingly, and that idea resonated with me. I think that's what actually ultimately drew me to logistics. I thought the concept was just very simplistic and just kind of this wonderful idea of if you save the dollar, you earn $1. But ultimately, we're -- today, we're going to talk about why that mentality is wrong because it's representative of this idea that supply chain is just a necessary evil and just the cost of doing business. Logistics and supply chain historically have seen a supportive role within the company. So the better you support your company, the more you invest into your supply chain, the higher the cost. And so with that mindset, companies were focused on reducing their expenses as much as possible while maintaining whatever minimum service standard they had as a company or that their company or a customer require. So the goal is to find the sweet spot where the incremental cost of better service met the incremental benefit of lower supply chain costs. So you wanted to maximize your surface service and minimize your costs, as easy as that. Now this idea of reducing costs has not gone anywhere. It's still a top priority for you, for your boss, for your boss's boss, for your customer. And when you think about cutting costs, what do you think of? RFPs, maybe reducing your transportation rates, utilizing containers or trucks better. In reality, only 20% of your supply chain costs can be affected by these types of tactical decisions. The other 80% is hardcoded in your network design, where you source from how many DCs you have, how you fulfill your customers' orders. And without changing that structure of your network, you're limited and your ability to affect your cost. Imagine for a second, if you bought a house when you started your career when you're 20 years old, you go out and you buy the best house you can afford. And then as life moves on, you add new rooms to your house. You get a raise at work, you add a new room. You get married, you add a room. You have a kid or 2, you got a few more rooms. It's not a house that I'm describing, which would be absolutely nuts. You have to go through your garage to get to your bonus room or maybe your kids room is located off the kitchen. In real life, when your family grows or your needs change, you go out and you buy it a different house. And if you never reconfigure the design of your supply chain network, your supply chain is going to look a lot like this house. Adding new facilities for new suppliers or transportation modes as your company grows is necessary, but if you don't perform regular routine network maintenance, you could be creating a lot of inefficiencies within your network design. Now the reason so much of our cost is tied up in this design element has to do with how all these nodes and entities of the supply chain are connected. So if you reduce your distribution expense, you're going to have to offset that with an increase somewhere else. So maybe your transportation spend goes up. If you reduce your transportation expense, maybe you hold more inventory, your inventory goes up. And that's reflective of the idea that a supply chain is a system. And so when we think about a system, I think the main idea of a system is that all the parts are connected, they're interdependent. There's some output or goals that you're trying to achieve, and no individual part can complete that task on its own. Any change you make to one part affects the other part's abilities to perform their role. With the case of the respiratory system, the parts work together to deliver oxygen to your brain and to keep you alive. Any long-term failure in any of these areas can be fatal. If you were to hold your breath, for instance, your lungs would continue to distribute oxygen, but eventually, they run out, and again, it'd be a fatal situation. So as we discussed supply chain activities, operating in silos, cost performance being measured in these vacuums, but that's not reality. The decisions made in each activity have a significant effect on the ability of these other activities you have to operate. So when we think about the design of our supply chain, we have to measure the impacts, the change you'll have on the full system, not just within the immediate reach of the particular activity we're trying to affect. Now fast forward to today's supply chain. Here's a simple representation of what supply chain is today. The complex web of transportation modes, touch points, customer fulfillment opportunities. Yes, freight moving forward. Yes, freight moving backwards, maybe side to side. And every year, there's new breakthroughs or new best practices that enter our industry. There's new disruptions that you have to navigate. And it's exhausting to keep up with all these new breakthroughs, but the goal is to create a more complex supply chain. The goal is to address the need to lower cost, but also the need to address some other, I guess, more complex functions of the supply chain. So as I mentioned earlier, supply chains were once just considered cost centers. But today, it's much more common to think about supply chain as a strategic enabler for your company to fulfill some set of goals. Maybe you want to deliver product to your customer more quickly so you implement some robust final mile delivery program. Or you're trying to be more resilient to respond to disruption. So you diversified sourcing or find additional inbound transportation options. Maybe you're trying to implement an omnichannel fulfillment strategy so you can reach more customers. You're going to have a hard time doing these things if you don't set up your supply chain to align [indiscernible]. And also, all these strategies can help set you apart from your competition. Some industries have more brand loyalty than others, but in every industry, we're seeing an increase in competition, which is putting more acute focus on customer satisfaction. Marketing has always been the piece of our business that gets credit for building brand loyalty. Their job is to tell a customer why they should choose our product or service, but supply chain gets to show them why they should choose the product or service. For me, I order almost everything through Amazon because I know it will be here tomorrow. In some cases, sooner. And that means fewer errands I have to run this weekend. To me, that convenience is invaluable. It trumps pretty much any other service consideration [ at the end ]. And Amazon has one of the most advanced network designs. They have used it to create incredible value for the company, and we're all feeling the impacts of Amazon. They impact just about every industry as a result. They change the way customers think and what their expectations are. So now to keep up with that change and expectation that increasingly competitive in volatile landscape, you have to design your supply chain with many more factors. So the first is speed. I just spoke to it. Amazon is changing the game. Every customer wants to have everything at their fingertips just a day away. So where are your customers located? That's where your product needs to be or at least accessible within a couple of days. To you that's going to mean smaller distribution centers, more distribution centers, forward stocking locations, enhanced delivery options. Agility and resilience. So disruptions are happening every day. If you're out of stock of an item or you have a truck of the long strike or for any number of reasons, you're unable to deliver to your customer within their expected lead time, what other options do they have? Are they loyal enough to wait for you to get the product to them? Things like holding extra safety stock, fulfilling orders out of another [ DC ], using another fulfillment option may be the only way to make sure we keep that sale. Sustainability. It's an increasingly important topic whether because of government regulations or just social pressure, there's an increasing need to consider the environmental impact of your business. Implementing consolidation programs is one way to reduce your footprint by reducing the number of loads, increasing optimization, making sure that you're targeting suppliers that use recycled goods or other green manufacturing solutions will help with this as well. These items have an additional cost, but it will help your company separate itself from the competition in the mind of your consumers. And of course, cost is still on here, still very important. You're not going to stay in business if you don't control your costs. And if you try to stay competitive in these areas I just mentioned to try and meet heightened customer expectations, but you don't have your freight in the right locations, you don't have the right inventory levels or an efficient way to get product to your customer, your costs are going to explode. And this is why the idea of supply chain design is such an important topic today. Supply chains are not a supporting role. They're not simply the cost of doing business, they're foundation of your company's value chain. And as we talked about our supply chain as a system earlier, now when we think about these value functions, they're connected as well. They're interdependent in the way that they support or engage with each other. There's a dozen ways you can affect the value and service level or lead time or cost. And each one of those is going to have a collateral impact on the other functions on this wheel. Some positive, some negative. And ultimately, it's not a zero-sum game. If you increase cost, you're not going to see the same reduction in lead time. You can see progressive movement, but ultimately, you can't perfect each and every item on the wheel. And so that's why it's important to understand your company, what your philosophy is and what are the relevant inputs to meet your goals. If I were to run a dollar store, I probably need to focus on cost as kind of the main driver of my decision-making. I have to be okay with holding higher levels of inventory, longer lead times. So I probably built large DCs that order full container loads from my suppliers. If I were to manufacture smartphones, I probably went shorter lead times and focus on better service level. So then I implement here parcel ship. I'd probably ship direct to my store to my customer. So we have to make sure as we're looking at our supply chain design that we're aligned with our company's philosophy as well because, again, there's so many different levers to pull. You're going to have to make decisions on what makes sense for your company, how you design your supply chain to align with the needs and the philosophy of your company. So now I'm going to hand it over to Shane, and he's going to talk to you a little bit more about the tools and the methods that you can use to support with this decision-making when evaluating your network design.

And let's pull up -- Raven, while we're here, let's pull up the results from the survey. I'd love to see kind of people's experience with supply chain design, where did we land with our audience. I see here. So I am seeing the survey results on a side panel. So we have about 50-50. So we have about half of the respondees that have done a -- conducted a supply chain design study in the last 1 to 3 years, and then we have the other half that have not done a particular supply chain design study. So interesting. I think the supply chain design that Hack talks through it, we're talking about formal design studies using advanced tools, but I actually believe that those that said they've not done a supply chain design study, probably have. But it's probably been simplistic. It's probably used Excel. You've made decisions on optimization using Excel and doing your own evaluation. And that's one level, and we're going to talk about different levels of analysis and kind of look at the way that Gartner talks about it. But before I get into looking and talking about analysis, I think that we should talk about Sherlock Holmes. And he has this great quote from the book of the Adventures of Copper Beaches where he said, "Data! Data! Data!.. I can't make bricks without clay." And the same thing can be said about doing analysis. So as we think about the foundational elements of doing analysis, of course, you have to have data. We can't do any analysis without data. And so Hack, if you go to the next slide, one of the things and one of the tools that we use within our team is a network diagram. So a lot of times, when we start any particular discussion about a project or our sales or account management teams may engage with clients about their supply chain, we really try to start with putting -- creating a visual representation of a customer supply chain, using in this example, a network diagram. And this will achieve a couple of things. First of which is, it creates a tremendous opportunity for us to talk about the different data landscape. So where is the data sources? Who are the different customers that may be shown on the network diagram? Where are -- is that data coming from? What are the systems are we talking about? Your ERP system? Are we talking about transportation management systems? Are we talking about your logistics service providers? And as you think about all the nodes in your supply chain and all the providers that are in [indiscernible], you probably have lots of different data sources and data locations that, that may be stored. Some of it may be in Excel spreadsheet on somebody's laptop. Some of it may be able to be queried against your ERP. Some of it maybe is stored on a monthly spend report, right? So those are all different data elements. And that really sets the foundation for any time we're going to do analysis, we can decide, well, what do we have in this data? What do maybe we not have? How do we connect the two maybe across different data sources? So ERP and order level data with transportation level data, there's going to be some data elements across both that we may need depending on the type of study, if we can blend them together to really get into product and transportation information. But this is a great tool to use to get a data landscape of your supply chain. Up on the top right, you'll see that there's a QR code. A couple of years ago, we wrote a white paper about sort of mastering supply chain data analysis. And we actually have a template that you can see that shown that it's a [ blank empathy ], you can download. And it's a great way to really start building out your own data landscape across the different areas in the supply chain. So you can look at the purchase order, the supply information or the delivery activities, and then you can go and start documenting along with building a network diagram, where are your data elements and how are they treated? Are they stored in a location? Are they just available at a location? And so as we think about a lot of what we're going to be talking about today and what I'm going to walk through is building analytics maturity. Really, the foundational item is starting to understand data locations and the quality of that data. But the benefits of a network diagram actually go beyond building a data landscape. And Hack, if you take us to the next question or next slide. So I'm giving an overview of a supply chain, right? So this is a written overview. It talks about the supply chain. They ship to stores and their e-commerce customers. It talks about the stores or serve from their own DCs, which are served by the -- to the customers in the Europe DC, which is in the Netherlands, and then there's one in the U.K. And then those same stores serve directly from the in-country DC. We also have web customers. There's lots of information about the customer supply chain. Now if I were to probably get inside each one of your head, you're going to have probably your own interpretation of what you just heard and what you just understood. Now the reality is that we all probably arrive at a different visual of the supply chain. And as we work with our stakeholders across multiple functions, Hack talked a little bit about the multiple groups that have competing interests, they all have different perspectives of what's going on in their supply chain. So building a network diagram, in this case, we have a visual diagram of all that same information that we just reviewed. And it allows the majority of humans, which have a tendency to learn with visual -- through visual learning to actually be able to see and understand that we can capture a tremendous amount of information about a supply chain using this network diagram. We can talk about the relative proportions of where the suppliers are. I think 40% in South China, 40% in Shanghai area. We can talk about the determination of when a consolidation center is used versus when it's a supplier direct. We can see all of the different stores that are serviced by the customer DCs. We can see the different DC locations and how the different forward flows and reverse flows, all correspond to each other. So this is another great way to align different stakeholders across your organization, give a visual representation of your baseline. And then as you guys talk about strategies, initiatives, clarifying what the sales group wants to do and the new go-to market. This provides a great playbook to have everybody singing from the same sheet of music about what's going on with this particular business or this particular strategy. So a really, really beneficial way to encapsulate the activities of the supply chain give you the ability to talk about data and analysis that can really set us up to be in a strong position to start building out the data landscape, to start to take analysis to the next level. So let's talk about data analysis. So I've pulled up here a visual from Gartner. Now anybody that's followed the Gartner, maybe you've seen this, you probably are more familiar with what they've used for many years, which is the analytics maturity matrix. And it was an arc of the different levels of maturity, and they've added a couple of elements on here that I thought were pretty interesting. So you have the same basically 4 maturity levels, and they've clarified these as the analysis types. So what type of analysis are you doing in your organization with the idea of saying, yes, we're Level 2 because we're doing both descriptive and diagnostic or we are a Level 4 or even a Level 5, because now they have this split out of where does the decision support come in versus decision automation, but they have these 4 different levels. What they've added is, they've added a couple of different elements related to decisions. And where do decisions come in? Is it human? How much human input is required? So as we think about on a more basic level, a descriptive -- of course, the analysis is giving you information, but you as the human, as the user really have to spend the majority of the -- take the majority of the information to consume it, to understand it, to try to make determinations around it to support a particular decision that then takes action within your supply chain. So as you go further and further down, the idea is that it's going to be more complex, the analysis type. So that's why you're higher on the maturity. It's going to move from hindsight, so looking backwards or looking at current activities. So then going to 4 sites. So from hindsight to inside giving you good information to then foresight about actually understanding what's coming at you and making decisions. And then you can see in the prescriptive, you have decision support and decision automation. And this is something that we'll talk a little bit about. I'll try to give a couple of examples, but of course, with machine learning and AI, they can build models that they train to actually take decisions within the supply chain, but there's certain limitations and certain things that we need to consider if we're going to be doing that. So this is a new take from Gartner that I think provides a little bit more information, but I'd like to walk through each one of these and kind of talk through it. And I'll give some perspective maybe of how we look at it and how we use it. So let's start with descriptive analytics. So this is what happened typical, what happened or what is happening. So what is happening? I took a screenshot from marine traffic. I think this was out of yesterday. This is the vessels, the container vessels around the ports of Charleston and Savannah, and we can see how many vessels are in queue, how many are coming in from the different ports and information. So it's telling us what's going on at this moment in those ports. We can make some determinations about it. It gives us some information to make a little bit more educational decisions, but it doesn't really tell us much more beyond that. Same thing about your report. So if you guys are getting spend reports, if you have your own KPIs about your warehouse performance and your activities, these are all descriptive analytics, telling us what is happening, what's going on. We can look at things over time. We can understand relative trends. But it doesn't really give us -- if we start to see a down trend, it doesn't tell us what happened, what was the cause is necessarily on that information of those lack of performance or where are we having a gap that we need to close between our expected performance versus what we're achieving. And so as we move into trying to mind and understand those opportunities, we would go into diagnostic analytics. So in this case, we can look, and I'm taking this an example of a customer and their outbound spend. So I may be a supply chain manager, and I am trying to figure out why my outbound spend has been trending up from my DCs, from my outbound flow. I did my network diagram. I've done a good job of organizing my data where I can have clean and consistent mode service levels. I've now put on some recurring analytics and visualizations of those analytics to see, in this case, histograms of my modes outbound. So I have LTLs and full truckloads, and I've broken them down into different weight buckets going from less than 100 pounds all the way up to 10,000 pounds. And I can see where my shipment profiles are fitting. What -- there's a couple of things that jump out of it. First of which would be the 3,000 to 5,000. So you can see that I have some full truckloads in LTLs. I know that my typical pivot weight is about 9,500 pounds. So as I started seeing this activity over the last week, it tells me that there may be some opportunities for me to revisit my PMS system protocols or my carrier allocations and load selection by the team because I should probably not be doing unless it's maybe a local shuttle move be doing a full truckload in shipments that are 3,000 to 5,000 pounds. Those are probably really a sweet spot for my LTL providers. Maybe there's some opportunity there. The other thing I can see is the 10,000. I probably should not be doing any unless I have some really impressive LTL rates on a specific line, I should probably not be seeing any LTL movements in that 10,000-pound category because that's my pivot point where full truckload is going to be better identified. And then on the very following side, there's only 2% of my shipments, but it looks like I'm not moving any of my parcel. I'm not moving all my parcel movements like I should be. It looks like I probably have about 2% of my shipments that are moving as LTL, although would probably be a better fit for my parcel provider. So this would be an example of, all right, I know what my spend is. Now I've started to put in some diagnostic analysis to say what may be a cause of that particular spend going up and have identified some opportunities. Another example might be model versus actual. So is my distribution network performing like it should. I have a 3 DC network. When I modeled it, I showed these on the left side. The locations in green right in the middle of those, those are where my DCs are. I expected to service the customers from the closest potential location, but in all reality, you can see that we have a lot of colors in reality that are spanning across different markets. And that tells me that I am not being as responsive to the customer as I should -- as I had expected to be, but I also am increasing my spend because I'm shipping from L.A. to Chicago or L.A. to North Carolina when I should have had that coming out of the facility in New Jersey or Arkansas. So this would be another view that I may look at on a recurring basis to see how is my planning team doing? How is our order allocation team doing? Are we having the right configurations in place to really make sure that we're making the best determinations of what customers get delivered from which DC, and I'm going to look at that over time. So that would be an example of a diagnostic analysis. So now we're going to talk about the next level predictive analysis. So this is where things get interesting. And I'm going to talk about it from sort of 2 different sides of the coin. So in this case, I'm going to talk about it from an optimization perspective. So a lot of times, we get asked to help optimize customers' networks. Well, we may not always use optimization analysis for an optimization technique or approach, we may use simulation along with the optimization. We may use both. And so when we talk about an actual technique of optimization, what we're really talking about is trying to find the most optimal results given a set of objectives, right? So what are we trying to achieve. But with a set of constraints, which I think is, to me, at least the way I interpret it and think about it is that constraints that those -- that's the most important characteristic that we're really building into a model. So we can optimize towards the goal of maybe of cost or the lowest cost and against competing trade-offs like lead time. And if you remember from Hack's view of you had cost and lead time and there was the sweet spot, right? We can optimize towards that, but maybe we have a constraint like carbon footprint, right, that we don't want to go above this amount of carbon footprint. And so that may mean that we are actually going to be a little bit more expensive, but we're not going to -- we're going to not go above this particular carbon footprint. And in this particular example, we did an optimization model for a customer where we looked at many variables, and we had many variables for a manufacturing customer where we looked at inbound logistics, manufacturing activities, the distribution activities and then outbound logistics. So if you'll see the simple network diagram on the left, we could see that we had supply coming in from suppliers as well as some of our component facilities that were offshore in Europe, in Slovakia, in Italy. We had 2 different manufacturing sites straddled on the coast in Oregon and in Pennsylvania, and we ultimately have these different customers that are in the middle of the U.S. in different parts. But one of the key things was, those facilities obviously don't have unlimited capacity. We can't push everything over to our site in Pennsylvania because they had a certain capacity on how much they can manufacture. So we had to put in a constraint of a manufacturing capacity limitation. And so in an optimization model, we have to look at all the possibilities of the inbound logistics, the manufacturing costs, the distribution costs, the outbound costs, lead times, but then add the constraint of what is the max output that they can provide so that we're not pushing all the volumes from the East Coast facility because most of the components are coming from Europe with some also coming in from Vietnam. So that would be an example of an optimization model. So the other side of the coin for predictive analysis is simulation. So I like to think about this as what will happen if we do x or y. And so we're taking the network characteristics of the network, historical data through the network, and we are simulating it across specific locations. So we may have done some demand analysis, the matter of center of gravity analysis to try to minimize customer distances from different COG locations, so from different potential locations. And then we're going to use simulation to actually see how those locations interact with our network and our profile. So where do we have volumes coming in versus where the customer is going out. So you have this trade-off of inbound costs and outbound costs. You can include distribution activities and distribution costs in those. And so we're going to look at how does Memphis interact with Dallas? How does Memphis interact with Norfolk? If we are in Norfolk, well, of course, we're going to have probably lower inbound costs, but the outbound costs are going to be higher because they follow -- the customer's profile follows the population of the U.S. So we know that there's many more customers that are going to be towards the middle of the U.S. that we can get to as opposed to just on the coast. So in simulation, we're basically running the activities against different locations to see how cost, service, lead times, all interact with one another. We also can use it to bring variability in or bring new dimensions in like forecast. So we might do it where we actually do an optimization model to find what are the -- what is the most optimal given our objectives and our constraints. But then we might bring in simulation to introduce variability. So what happens if the volumes go up by 20% during this time period or the shipment profiles go down by 20%. So we can test different configurations, different scenarios by running simulations with additional characteristics for additional variability. So oftentimes, when you're doing supply chain design work or where you're doing more predictive analytics, there's actually this combination of things that you can use, and it's just different different approaches and setup of a particular model. So oftentimes, we're optimizing a customer's network, actually doing simulation, but we may use simulation on top of optimization. So they kind of work hand-in-hand. All right. So let's go to prescriptive. So this is our last one. This is -- can have lots of different dimensions. I'm going to talk about regression analysis. Regression analysis is, I think, a statistical technique that's used probably most consistently. And I sort of reframe this of what should I do given what's likely to happen with the key piece being what is likely to happen? And so we have -- if you look at the chart on the left where it says regression analysis, we have a correlation chart, and we have a dependent variable with independent variables. And so the dependent variable can be, what is the thing that you're most interested in? And then the independent variables are, what are the things that you think might cause a particular problem or challenge? And the regression analysis is going to really separate the signal from the noise to help us understand where is there a strong correlation between these 2 variables that is actually causing one to do the other. So an example of that I can give that our team has done is lead times. So lead times against maybe a given mode or a given lane. Maybe we're tracking lead times and there are reason codes that are associated to why this particular shipment was not on time. So maybe we have a given KPI and we have on time or not on time. And so we can start to use a regression analysis to start to understand how much correlation was their when this activity happened that it created the product to not arrive on time. And what the bottom chart is saying is that when we have something that starts in the lower left and goes up to the right, the farther the right it is and the higher, essentially the more confident that you can be that there is a correlation. So when we have a product that is shipping from Korea into the U.S. on this port with this day with this carrier, we can run statistical analysis to identify that there is a high likelihood that this shipment may be delayed when it gets to this particular port. And so what that can then turn into is that you have started to understand the trend that you can get out in front of. And you have a high confidence because you've done a statistical analysis of the past to know that there is going to be likely an issue on that lane. And so if you have a high understanding of that, the next level would be to actually, well, take an action. You as a human, if it alerts you and you do it, but to actually be able to get to the spot where maybe the -- your system is taking an action and you create an algorithm to say, when we meet this criteria where there's a high correlation, I probably need to start either alerting your customer service team. I need to start by bringing in additional products and increasing the purchase order to account for this product is likely going to be delayed. So that's an example of being prescriptive, and these items exist. These opportunities exist all across the supply chain. And Hack talked about Amazon, who I think has really been probably at the forefront of this. They probably have thousands of team members that are creating these little models or various aspects in their -- the movements of their supply chain. Route optimization, inventory levels, inventory and allocation. So a couple of examples that I have is a predictive ETA. So Expeditors last year launched a -- I think it's called Enhanced Container Tracking. And so it's taking multiple data inputs from GPS signals, so real-time data that they're getting. They're getting information from carriers, and they have historical data. What we found or what the team found is, they've looked at container tracking over a number of years and probably many of you can relate to this is that a carrier may be saying one thing, the terminal may be saying another thing and there are all these conflicting pieces of information. And so how do we separate the signal from the noise on what is actually most accurate and what you can expect? And if you can expect that it's going to be there a couple of days earlier or a couple of days later, then you have the opportunity to start making more proactive decisions about labor and inventory levels and customer service activities and consolidation opportunities to bring more products in. So it opens up whole world of different ways to think about optimizing your supply chain and getting that incremental benefit that Hack talked about in the very beginning. So I put a QR code there. If you want to watch, I think there was a blog and a video that you could pull up that kind of describe that in a little bit more detail. But this is possible for other aspects of the supply chain. So for store allocation, right? If we see a spike in demand and we have activities of where the inventory is across different areas of the supply chain, we could maybe do some of the reallocations automatically from DCs that maybe aren't naturally close to one another, distribute from another customer DC. Or if we're using point of sales data, and we can be closer to sensing the demand, that might create an algorithm where we're understanding where is our current inventory, what are we expecting for next week's demand given historical activity that we've seen and a supply lead time. Well, we better get out in front of this because in 3 or 4 weeks, we're expecting that this inventory is likely going to be depleted, and we're not going to have it in stock. And so it's getting smarter to help the planning teams be able to make decisions, changing the way that they look at demand on a more granular level as opposed to maybe rolling it up and looking at things on a 4-week or a 6-week kind of rolling time frame. So these sorts of opportunities are all around in the supply chain, but it's all about starting with the basics of the data and being able to build the foundations and incorporate this over time. And Hack, I think the last slide is that just to help make you comfortable, this is a journey. Amazon didn't build these capabilities overnight. It's a journey, and I'm going to turn it over to Melissa that I think is going to talk about some of the key ingredients to help you on this particular journey.

Thanks, Shane. Appreciate it. So let's review what we consider the key ingredients for success in supply chain design, and I have a few overarching comments to make as introduction. And that is, a better design starts with knowledge i.e., knowing where you're headed. A better design also follows principles in the process. A better design can result when you have a keen focus in on the business problems or opportunities that you're trying to solve. And really what I'm talking about is the strategy, right? The strategy provides the overarching directives for the supply chain organization. But you know what, it's the people, the process and the technology that really turn the strategy into action. So strategy is for approach hypothesis, whatever label you want to give it, that really is the foundational starting point here, one of the key ingredients. One of the next key gradients, it has to do with skill sets, and it's having those knowledge experts and decision makers. Those are the keys to any type of -- keys to success for any type of project, and certainly, sponsorship too, because sponsorship not only sets the expectations, it also gives purpose and direction to the team and certainly, governance. It's key to have those clear owners, who -- to actually ensure that the functional stakeholders and other partners remain on task as well as updated on a routine basis. So what we're talking about here is business subject matter experts. We bring in the supply chain expertise, the logistics, the trade compliance. Functional SMEs, sales, marketing, customer service and let's not forget the data SMEs as well as the analytical and the modeling talent. So by having all these subject matter experts, there are 2 other characteristics of say, like a successful team, a very rounded set of skill sets. And those characteristics are -- the individuals are problem solvers as well as those individuals are change agents in amongst themselves and band together as a group. And you know what, we can't talk about supply chain design and projects unless we talk about data, and Shane kicked off his section talking about data. And you know what, we have to -- I just want to give a nod to the fact that it can really be an internal challenge to overcome. We face this day in and day out with every project that we face. So why is it so elusive? Well, I don't know, some kind of things are multiple systems of record, ongoing system migrations and sometimes a lack of overall data government to name a few. But you know what, large-scale transformations require core inputs like data. So the proverbial garbage in, garbage out is definitely in play. So what Shane was talking about this data landscape, this is why we advocate on identifying what we call your data deserts, where you have pockets or gaps and closing them because these are the key enablers for the future. And moving on to the process, you know what, the supply chain design project doesn't require any special type of magic. However, it does require following a diligent and methodical project management process. And you know what, this is very powerful. And what it does is, it places structure around the future of the supply chain and how we can calculate what would happen if certain parameters were changed. And then finally, on the technology side. Well, we need tools that are capable of quantifying and balancing the complex constraints that exist in our global and interconnected supply chains, given some high expectations or high objectives of a particular study. This is a reality that is driving many companies and our customers to seek more robust solutions and why the supply chain design space has really grown. It's the tools and the computing power. It's all better now than it's ever been. And so therefore, and couple that with disruptions and the volatility, this is the focus and the capability. And now we have the ability for the solutions to be brought to even a higher level. But let me just say one thing about technology, and I'm going to elaborate on this in just a minute. We advocate that spreadsheet technology is really inadequate for rendering complex systems like supply chains. So if we can move forward a little bit and take a little bit deeper dive into the project life cycle as well as the technology. And I see that we're bumping up against the top of the hour here, so why don't I just hit the high points here we follow and apply in methodology, and we do it based upon the scientific method. And once we define what a project is, next phase is to go into feasibility. You know what, basically, what we're trying to assess whether or not is the data that's collected sufficient for us to move on with the project, and that's why we see that first black triangle. If it isn't, then we need to go back and either abandon the project or collect more data or make robust assumptions. Then we move into the baseline phase, then into the due diligence phase. And that -- the due diligence as well as the detailed design phase is around taking that baseline and comparing it to a slight number of scenarios, and then those ultimate models then simulate the impact of -- impact to the network and the flows by utilizing that data that we had in the baseline. But you know what, this is just a onetime project, and a onetime project and thinking about the volatility that we have in our supply chain here today really do require a different approach. So Hack, I'm going to ask you to go to the next slide for the sake of time here so that we can talk a bit more about the technology itself. So granted, we -- I made a statement about Excel approaches being insufficient, but much of what Shane and Hack, and I have been talking about today is a onetime project. And while that is the current capability, we want to give you a perspective of what a best-in-class capability is. And that is when a model is fed constantly with continuous data and it allows companies to ongoing measure what's happening today and the design for the future. And you know what, that is one of the key components to the digital twin difference is, you move from a periodic a onetime only into a continues approach, and therefore, you shorten the time to [indiscernible]. So again, for the sake of time, Hack, can we move forward, please. So a digital twin technology is -- our living model system is built from the digital twin technology. And because we're running up to the top of the hour, I wanted to give you a little bit of a tease into the deeper dive that we're going to do in about 2 weeks, in which [ Christie Meyer ] is going to take us through Expeditors authoring of the digital twin and the practices that it uses here in the supply chain design space. So final thoughts here, elevate your analytics, move beyond static spreadsheets and invest in tools and capability that enable a nimble culture. And really the best-in-class approach here is in the digital twin, being able to do continuous scenario modeling, and therefore, empowering agile decision making. So we certainly hope that you join us here in 2 weeks, and I think we do have some more resources on the next slide that you might be interested in with our horizon breed. And if you click one more time, Hack, both the Horizon [ breed ] and the biweekly newsletter. So thanks, everyone, for your time. I'm sorry, we ran out of time for questions. If you posted any, we will be sure to get back with you. Back to our host. Thanks.

Thank you so much, Melissa, and Shane and Hack. This is an excellent seminar, so thank you so much. Actually, we -- for anyone who needs to drop off, we totally understand that we're above the top of the hour. But I did want to go ahead and just list there are a couple of questions just for the sake of the team. And I'll just go ahead and mention those. And the first one was, just given the volatility with the rates in the market, how do you handle this with your modeling? Who wants to take...

Sure. Well, yes, I think the rates -- they've been very volatile, of course, to levels we've never seen in the past. I would say that the thing that we're really looking at is trying to -- well, we could do a couple of things. So trying to find what like agreement on where we should be looking, on where we think the market may go, but this is a great example of where a sensitivity analysis could come in. So what happens if the rates go down 20%? Are these locations still optimal? What happens if rates go back up 20%? And you can start to look at it from those different dimensions to see how the network performs under different rating models. I think it really shows the power of being able to build something that you can test in multiple ways and really tees up for the living model and what we're going to be getting into in the next session being able to revisit on a frequent basis.

Okay. Thank you. That's a very relevant question. So for sure. And the very last one, the rest, we will go offline for any others that had not been already addressed, but this one, I don't know if it was addressed because I didn't hear it. Do you see any trending regarding typical failures or pitfalls that should be upweighted when you're performing supply chain design studies? You don't mind to think about that one?

Yes. Good question. I usually like to spend it more towards like best practices. We talked about the skill sets, right? The different skill set, having a clear objective, the sponsorship and following a robust process. I realized we weren't able to get into many of the details here today, but those are the key essentials. That's why I said, there are key ingredients to success. Shane or Hack, do you have other ones that come to top of mind?

I can make a joke, Melissa, but I'll stick to -- yes, your key ingredients for success.

And Hack, mind to give back just one slide. If you guys missed this, our next seminar that's coming up, it's going to be about the living model. And -- so here's the QR code you will receive for registration for this as well on September 27. The time is 11 a.m. Eastern. So we will have that for you in just a couple of weeks. So thank you so much, our panelists, for speaking today, and thank you for everyone who's joined. We are looking forward to having you on our next webinar. Have a great day, guys.

Thanks, everybody.

Thank you.