Reuters Events | Supply Chain & Logistics Business Intelligence

Putting tech at the heart of XPO’s operations is helping to keep a logistics giant agile amid a global pandemic

How does a company with a global footprint remain nimble and adapt to the unprecedented times we are in?

Alex Hadwick, Editor-in-Chief for Supply Chains, Reuters Events, sat down with the Chief Information Officer of logistics giant XPO, Mario Harik, to see how the company is evolving its technology platform and responding to COVID-19.

This interview has been edited and condensed.

Alex: Where have the shifts in demand been apparent as a result of the pandemic and where have gaps occurred in supply chains?

Mario: A lot of what we have seen has been part of a 15 year-long consumer story, and that's the rise of e commerce. We did a consumer survey and what we found was 94% of consumers were more likely to buy something online than they would have pre-COVID.

The product categories of things that they were buying online shifted as well.

Before COVID-19, it was most likely that a person would buy, says groceries from a physical store, as opposed to buying online. Similarly, with personal care type products or healthcare type products, apparel, fashion-type products, these were all things that people were more likely than not to buy in store versus categories, say like electronics. We've seen the demand for these type of products increase a lot through the pandemic.

Now, our customers were faced with supply chains that needed to be nimble and needed to be agile. They needed to cater for changing demand from one side, and then make sure that the supply chain can react very quickly to respond to that demand overall.

Our focus has always been on how do you manage those flexes ups and downs from a customer demand perspective? That need for speed, that need for visibility, that need for agility, and making sure that we can fulfil that on behalf of our customers.

That was one of the keys when I think about the technology being applied to the supply chain and how you optimise it.

Our focus has always been on how do you manage those flexes ups and downs from a customer demand perspective? That need for speed, that need for visibility, that need for agility, and making sure that we can fulfil that on behalf of our customers.

In terms of specifically looking at COVID, there was also a shift in how companies operate. So, I'll give you an example. In the UK, we delivered more than 200,000 masks and face shields to hospitals in the UK with the NHS. This is an example where you have much faster demand windows. It’s not like you have a steady supply chain and the product is coming in from one point to another one, and you can pre-plan it. You're much more reactive. You have to move quickly. You have to provide solutions that are fast.

With our healthcare customers, suddenly they saw an increase of 40% more deliveries to hospitals through the pandemic

Similarly, here in the US one of our customers is Ford Motor Company, the car company. They switched a number of their manufacturing facilities to actually make PPE. What we what we did is we helped them move hundreds of shipments of visors, N95 masks, and gowns to hospitals, fire departments, police stations and assisted living centres across the US. You go from potentially needing spare parts at a facility, to now you're having masks go to the hospital.

So, you're shifting the nature of your supply chain and the type of product that you are moving.

Similarly, with our healthcare customers, suddenly they saw an increase of 40% more deliveries to hospitals through the pandemic. These were critical supplies: masks, gloves, gowns, IV components, and surgical kits that really were needed as part as part of the pandemic.

When you think about supply chains, and the technology applied to it, and especially in the context of the pandemic (but also we see it beyond the pandemic), it was around the need for agility: Changing how your supply chain operates; changing where the product is being sourced from; and what type of products have high demand or low demand.

Our technology platform has all the components that allow us to cater for that for our customers and really support them through what was a very different world over the last couple of months with the pandemic.

Alex: How has your technology platform evolved in 2020 and what roles have Application Programming Interfaces (APIs) played?

Mario: As a supply chain company, technology has been part of our DNA since the onset of the company and I've been with the company, by the way, since we got started back in 2011. My team got to build the entire technology stack  from the ground up.

In many ways as a supply chain business, we're in the data business

If I break it down, we look at technology in three broad categories. The first one is around data science and machine learning and AI.

In many ways as a supply chain business, we're in the data business. We collect data from customers about orders, where things need to go from point A to point B, or from consumers. We collect data from our trucks and around how they operate. We collect data from our dock workers, from our warehouse workers, from robot robotic solutions. We use that data to be able to provide a better service for our customers and to optimize how we operate the supply chain.

By leveraging our technology and our machine learning capabilities we can make sure that we can organize the products within the supply chain as efficiently as possible

One of our capabilities in machine learning is being able to forecast where consumer or customer demand is going to come from, and be able to position inventory on behalf of the customer in the right places to make sure that we can meet the consumer at a point of  heightened demand.

Imagine if a customer is selling cell phones in this potential scenario. We know that this blue cell phone is more likely to be bought in the northeast in the US, say in Boston. We can forecast with a high degree of accuracy how many units of that product are needed, say in the northeast warehouse for that customer that we're working for. By leveraging our technology and our machine learning capabilities we can make sure that we can organize the products within the supply chain as efficiently as possible.

Other examples are how we operate our routes in trucking and how we use the data that we collect to better optimize the routes, so they are reasonable. In the context of COVID, you need to be able to add dwell time on a per stop basis, for example, because you need more time with the social distancing, with PPE, with how the freight is being handled, to ensure it's being done in a safe way.

We can add in these things to optimise our routes very efficiently internally, and take that all the way to giving visibility to the customer on that data.

For example, we launched the COVID-19 dashboard. This allows our customers to see where COVID has shut down infrastructure, such as ports or municipalities and how the flow of freight is impacted within a certain geography. All of that is driven by the power of data and being able to collect that data and then optimise the business accordingly.

The second big area of technology is at our freight marketplace.

Today we have more than 60,000 carriers on the platform

We believe that the world of transportation is going to get automated over time, to where a customer whenever they want to move goods from point A to point B, they just say ‘I want to move these goods’ and our system, XPO Connect, can find the best mode of transportation for them. That can be us as a capacity provider, a company who has trucks or a transportation network, or using third party carriers to find the best fit for that particular customer, all electronically.

To give you an example, today, our carriers can use apps like the Drive XPO app on their cell phones to be able to buy a shipment now with a bid. We will support full end-to-end API integration for the public shortly, as the freight marketplace is all around giving a lot of self-service automated tools for customers, and giving them end-to-end visibility on how the supply chains operating.

Today we have more than 60,000 carriers on the platform.

When it comes to APIs, all of our systems, because they are proprietary, are API-enabled. What we get to see with larger customers is their own systems as well. We do full end-to-end integration with their systems, where they can either quote, or book a certain shipment, or track a certain shipment, all electronically, so the system is doing that with us, as opposed to having a person log into a system.  

When you think about e-commerce, the need for speed, then the need for being able to process things at volume becomes incredibly important. Automation is a big answer for making that happen

Similarly, on API's in our supply chain business, we are the largest e-commerce fulfiller in Western Europe and one of the largest in North America, so you can imagine the level of integration you need with the customer. When somebody goes on their website and buys something, we need to have that two-way sync of orders flowing through our systems, and then being able to give back inventory availability. When somebody goes to use our services to do that e-com fulfilment, that customer would see that that there's no inventory for that product, or they can see that this product is going to get shipped on a certain date, or they can see once they buy it, what's the estimated delivery time based on the fulfilment cycle for that product. All of that happens through API integration between the customer systems and e-commerce or any RP type system with our warehouse management/order management type systems that allow us to fulfil on those on those promises.

The third category is around robotics and automation.

The cobot can autonomously verify the product and go by itself to the to the packaging station. That makes a colleague in the warehouse twice as efficient as they would otherwise be

When you think about e-commerce, the need for speed, then the need for being able to process things at volume becomes incredibly important. Automation is a big answer for making that happen.

These are technologies like ‘cobots’, collaborative robots, for example. These are autonomous robots that work with our human colleagues in the warehouses. The cobot can autonomously verify the product and go by itself to the to the packaging station. That makes a colleague in the warehouse twice as efficient as they would otherwise be.

Same thing with goods to person robotics, where you're going to get a certain cobot that links a shelf of product to a stationary picker. And that helps a lot, by the way, with social distancing, because robots don't get sick, and they don't get the virus. So, it helps a lot with making sure that our employees are safe as well with a shelf of products coming to the stationery picker.

The automation space is about how do we make the operation of a given warehouse much more efficient, so we can support higher throughputs, we can support higher peaks, lower valleys and be able to respond to the customer demand on those pieces?

These are the three broad categories. One is data science and machine learning, one is our freight marketplace, XPO Connect, and then one is our warehouse automation. Large scale fulfilment is increasingly highly automated, and that's part of the reason why we spend roughly $500 million a year on technology, because we believe that's a core part of optimising and making supply chains efficient.

Alex: With a ‘black swan’ event such as this, algorithmic models trained on prior data may not be able to fully reflect the evolution of this crisis and anticipate the needed response. How do you cope with that and make systems robust?

Mario: Typically, machine learning and AI uses history to be able to predict the future. And obviously, when you have variability in the present, then the machine learning algorithms take a bit of time to catch up.

Now, the good news for us is we honed our algorithms to have really three levels of components.

One is that they look at short term history as being a big driver of the output. Then they get dampened by medium-term history. Then they condense even further through long-term history.

So, effectively, you're using three broad categories of data based on short-term history, medium-term history, and the long-term history.

Our algorithms over time had been honed to react to short term changes very effectively and very quickly

That helped us a lot and how we can react, so I'm going to give you an example of one of our platforms, which is called XPO Smart.

You can imagine that if the customer comes to you in a given day and says ‘I expect to have 55,000 orders today for our customers, so we staff up to make sure that we have enough people to fulfil the needs of 55,000 orders for that customer. Now if the orders come in at 65,000 or 75,000, we are short staffed, which makes service a problem because we can no longer hit the targets.

Therefore, to cope with this, our algorithms over time had been honed to react to short term changes very effectively and very quickly.

COVID pushed our employees in the field to use our platforms more effectively and more frequently, so they can react. Our platform allows us to come back to a customer when they say, ‘we expect  55,000 orders a day’, and then we go back to them and we said we tell them you're going to get 62,540 orders today, not 55,000. This is what our algorithms are saying and we accordingly change that piece.

The world of supply chains doesn't operate in months and years, it actually operates in hours and days

This is only one example, obviously, of how we look at data and how quickly it can react to volume fluctuations, because the world of supply chains doesn't operate in months and years, it actually operates in hours and days. Knowing that allowed us to react very quickly in the context of the pandemic. So when you see trends for a certain product, and they call them high-velocity SKUs, these are products that moving pretty quickly through a given facility or through the supply chain, we were able to see those trends much quicker. Them we can make sure that our technology, our people, all the processes of how we operate the supply chain, we're moving just as quickly to allow for that.