McKinsey on making your supply chain resilient: AI, tariffs, and digitalization

What do you get when McKinsey supply chain experts team up with Celonis technology leaders? Straight talk, practical insights, and a vision for how businesses can stay resilient in turbulent times.

In this special McKinsey x Celonis video series, McKinsey experts and Celonis leaders discuss the key challenges that supply chains face today:

• How AI is reshaping supply chains (and why humans still matter)
• How tariffs can become a competitive advantage, not just a compliance headache
• Why digitalization and data quality can make or break resilience

Scroll on to read the full interviews:

Episode 1: Augmenting with AI

Speakers:
Julian Fischer – Partner, McKinsey & Company
Sayali Chavan – Celonis

Q: AI is the top priority for everyone. But do you see AI as a cure-all for supply chains?

Julian Fischer (McKinsey): Yes and no. Yes, because it's one of the largest disruptions we've seen recently, and it's much more than just a buzzword. We’ve had many of those in supply chains over the last years, blockchain and other topics, but they haven't really changed the rules of the game.

AI is really comparable to the rise and the implementation of the containers back in the 1970s, which changed the way that we transport goods and goods flow globally.

But the big difference is back then, it took 20 years for those containers to actually be standardized across the globe. This time it's going to go much faster. But is it going to be a cure-all? No, because there's still many areas where human touch is needed.
To give you one example, with a client of mine, we looked into demand planning and demand forecasting. Of course, advanced analytics can help tremendously to boost the quality of the forecast, using larger data sets and internal and external factors. But what we also found is that the best results were always achieved when a human had the last say and added in that extra bit of magic, market experience, and customer discussions to the mix.

That for me is the most powerful combination: AI to augment the system and then humans to really focus on the highest value-adding tasks to make it distinctive in terms of results.

Q: What should supply chain teams using AI today do to stay competitive and be ready for transformation?

Julian: AI can augment supply chains end-to-end. It starts on the forecasting side, leveraging AI and machine learning to better predict customer demand. It then goes into the entire production process, where, for example, production scheduling in a complex network, allocation of volumes to sites, sequencing of production, all this mathematical optimization and sequencing can really be augmented with AI.

It then goes into logistics and the transport: how do I fill my truck? What's the best sequence of packaging certain parcels or boxes into a vehicle? It also goes into routing towards the customer, and track and trace. So, by now, it has augmented the supply chain end to end. And if you look at the players in the industry that are most competitive, that's what they're doing. They're looking at the full potential.

Q: When we talk about implementing AI, we also need the right talent. What are the challenges for recruiting and training?

Julian: That's a huge question. Most clients that I speak with, when it comes to AI, it's not so much a question about the tech stack anymore, but talent. Talent is the key enabler. So how do I get the right talent in place?

Our experience in running the first pilots with real agents for supply chain is that you need to train them, and think about them like a new hire, a new team member. They come on board, you build the capabilities, you teach them – and you also try to teach them all the things that are not in the data. All that contextual knowledge, that tribal knowledge that's in the organization. And that takes time, but it also takes trust.

If you think about the average supply chain planner that I interact with, usually somebody in their fifties, they’ve been around the company for at least 30 years, working in Excel all their lives. They're excellent firefighters, they love their job, and transforming that capability profile into a new generation. How do I excite these colleagues to take on the task of learning around AI, of augmenting their work, of rethinking how they do things is a huge undertaking that can also only be done jointly between HR and supply chain. The two have to come together to do this. One of them by themselves cannot succeed alone.

I had a recent discussion with a planner who also said, "Look, why should I teach an agent that's going to replace me tomorrow? That's not in my interest." I explained that this is the wrong way to think about it. That AI will augment, will support, and also take away a lot of the tasks that are non-value-adding. Tasks are repetitive, that are not fun – and not why somebody chose a profession in supply chain.

Q: How can companies use AI for proactive risk management across the enterprise?

Julian: The level of transparency on risk has increased tremendously since COVID, and companies have invested very heavily into tracking and tracing all of their flows. This includes customer signals, but also supplier performance, and output of production sites.

What I see as the biggest potential is linking these together. If, for example, a supplier is disrupted, if a supplier's manufacturing site is affected by disaster in any way, what are the implications for myself? What does it mean for me? Tariffs are another great example. If a certain tariff scenario comes to be, what does it mean for me, for my flows and for my suppliers, and their suppliers? So it's linking those parts together to come to one conclusion on the implications, and doing that very quickly instead of taking weeks to analyze and crunch data.

Q: Could you share a story that really shows AI’s power in supply chains?

Julian: One of the areas I'm most passionate about is using AI for scheduling and optimization. I have so much respect for the planners who, manually, from their experience, are able to take a complex manufacturing network and turn it all into this production sequence which is feasible, which optimizes asset utilization, and delivers on time and full.

But if you walk through a modern manufacturing facility, the steps and the interrelations between different assets are so complex, doing that manually is nearly mission impossible.

Having AI to facilitate a real optimization with transparency and capacities on bottlenecks, on routings, is a fantastic application. And that’s when this whole notion of a digital twin of a manufacturing facility for scheduling really comes in – and is extremely powerful.

Episode 2: Transparency on tariffs

Speakers:
Marilu Destino – Senior Engagement Manager, Supply Chain Strategy & Operations, McKinsey & Company
Julian Fischer – Partner, McKinsey & Company
Nele Rittweger – Celonis

Q: What can be most useful when managing the impact of tariffs?

Marilu Destino (McKinsey): First, it's important to start mapping what are your vulnerabilities across the value chain, and really create transparency on the risk you have through tiers and mapping of suppliers. Secondly, it's important to think about, "How do I do scenario planning?" So, adopting tools to model different scenarios, understanding the impact of the risk on the business.

Thirdly, think about setting up a geopolitical nerve center, which helps bring together cross-functional teams, with predefined roles and responsibilities and decision-making power to really define the short-term and the long-term actions which are needed to thrive in the long-term.

Gen AI is very powerful in the sense that it's able to encapsulate this environment, and also, with the natural language interface, allows the planner to really interpret and understand the recommendations and the decisions.

The next level of this will come with agentic AI, where we are moving not only from recommendations for decision making, to really planning the recommendations back in the system.

Q: We've had a lot of disruption, a lot of black swan events over the past five years. Because of that, would you say that supply chain leaders are now better at adapting to disruptions?

Marilu: Yes, there has been a step change in terms of risk and resilience levers across many of the companies we've interviewed in our supply chain pulse survey. For example, for the first time since 2020, companies are relying a lot less on inventory as a short-term fix. Between 60 and 73% of companies are actually adopting dual sourcing and regionalization to become more resilient. And 67% of respondents are in the middle of advanced planning system implementation, which really enhances their technical capabilities in planning.

However, there are still quite a lot of actions that are needed to improve risk identification. In fact, the visibility into Tier 2 suppliers has decreased by seven percentage points since 2023. And only 33% of companies are actually setting up an early warning system, which helps integrate external and internal data in real time.

Q: When it comes to tariff response, how can supply chain planning help?

Marilu: It's really important to think about tariffs more as a strategic topic rather than a compliance issue. Companies that do that are better positioned to actually turn tariff challenges into a competitive advantage.

It’s typically beneficial to set up a geopolitical nerve center. This is a hub where cross-functional teams come together to monitor and analyze risk, and also define key next steps, both from a short-term perspective, and a long-term structural standpoint.

Q: What has been the impact of tariffs on logistics strategy?

Marilu: There has been a ripple effect due to stockpiling. This has impacted logistic capacity, spot rates, and regional trucking availability. So when there is a tariff which is announced, or actually comes into play, typically there is a surge in freight demand.

That has also impacted the way logistic operators tend to prioritize speed over cost, with some of the shipping modes switching from ocean freight to air freight.

This in turn has caused assets and drivers to be drawn to congested ports, lack of trucking availability, and also bottlenecks in warehousing, and increasing warehousing fees.

So for the long-term, it's important to think about moving from fixed freight contracts to more real-time capacity-driven models. Think about whether it's beneficial to regionalize or nearshore your supply chain. And also create that transparency on your TMS and freight visibility.

Q: What are you seeing as an opportunity for supply chain when it comes to tariff management?

Julian Fischer (McKinsey): The key thing that many organizations struggle with is, "How do I create transparency? What are the implications? And, how do I translate transparency into action?" That first step of transparency is a very important one, because when it comes to tariffs, you really have to understand your supply chain. So, where are your suppliers? Where do they source from? What does it mean for your products? And, what are the implications of a potential tariff scenario?

The changes that are coming are happening so quickly that you have to be able to rerun your scenarios and say, "If it's 10, 20, 50, more, what will happen then? Will my suppliers still deliver? Do I need to change my footprint?

Q: So this means we really need to understand our processes better first?

Julian: Absolutely. Processes: how does the supply chain flow, and also how do you work across functions. Many risks at the moment don't only affect one plant or one organization – it really is end-to-end.

One of the instincts that many organizations have, particularly in Germany, where I'm from, is to put a task force on something and say, "There's a big problem. I can't really solve it right now. Let's put a task force on it."

But given the fact that disruption is not going away at the moment, and it's just a constant way of working, a task force is just a band-aid that you put. You also need to change the fundamental underlying processes, the responsibilities, the tech enablement to be able to deal with these crises. And this is something which was a painful process for many organizations to learn and say, "We really need to change the way of working. We cannot just put a task force on top."

Q: What do the shifting logistics around tariffs mean for regionalization?

Julian: Regionalization is a huge topic at the moment in many different supply chains. The last, probably, two decades have been all around building global supply chains, sourcing from best-cost country locations, and then supplying the world, with all the free trade agreements and movement of goods supporting this.

Now, with higher volatility, with a lot of the disruptions we've also seen, that is being reversed, and a lot of industries are thinking about,"How do I localize more? How do I actually source from within the region? How do I also reduce the lead time?"

If an incoming material to a manufacturing plant here in Germany is on a ship from China for three months, those three months, you're locked into what you have coming towards you. So if your demand spikes or plummets, you cannot react anymore. That means that you have to be very certain of what you will need in the future, and that certainty, in many industries at the moment, is just gone.

Q: With all that pressure, it can be hard to know where to focus. So where should supply chain leaders prioritize resources?

Julian: You can get lost so easily when it comes to risk management and all the volatility at the moment. One of my clients, in the early days of their risk management journey, created a report on the most relevant risks, and he got a thousand of them.

One of the most important filters is always, will it affect my customer? Or will it affect my patient, if I'm in life science? And, does it endanger an order that I have? For each supply chain, independent of the industry, it's the customer success and consumer happiness that will be key.

Episode 3: Drive for digitalization

Speakers:
Julian Fischer – Partner, McKinsey & Company
Sayali Chavan – Celonis

Q: How are teams using AI effectively to get visibility and take action on supply chain issues?

Julian Fischer (McKinsey): It all starts with data and transparency. The big first area where we see huge shifts over recent years is on the inbound side – the ability to track and trace incoming materials, whether they’re on a container or somewhere in a ship, on a truck, on rail. Companies are using that transparency to predict what the effect will be on the rest of the supply chain.

Take a simple example: if I have a container that, due to the Red Sea crisis, has to go around Africa, what does that mean for my customer order in the system? Do I need to reshuffle and resequence my production to be able to still deliver on time?

If you want to do that in Excel, on a piece of paper, it doesn't work anymore. So you need AI to combine data elements, to work across functions, and improve processes end to end, to save that customer order and still be on time.

Q: There’s so much tech out there. Many companies are using it, but they’re still struggling with process visibility. Why does that happen?

Julian: Process visibility is indeed key. Because if you don't understand your processes, how will you increase your resilience? How will you actually remain resilient in these turbulent times? And why have folks not really understood their processes yet? It’s because often, processes are hidden, and we have to make them transparent first.

When I started my career in supply chain many years ago, in automotive, we used to map processes on huge walls of brown paper and try to get the essence of it. But you never succeed, of course, because they are too complex, and then you have these huge things and you don't even know what to do with them anymore. And the truth is, many clients that I serve are still at that stage at the moment. They do not yet have the right tech infrastructure to create process transparency. Very often, their data and their IT tech stack doesn't enable them to build this process transparency end-to-end and across systems.

But now, with technology, the possibilities are endless when it comes to transparency and processes. Getting that right, making that shift both from a tech stack, but also from a mindset, behavior, or a change management point of view, is really key.

I spend day in and day out with clients, with executives on the supply chain side, and what they all have in common, independent of the industry, independent of the region, is data quality. And everybody says, "Oh, our data quality is so poor." And then I ask, "What are you doing on it?". “Not a lot,” right?

For me, data quality is one of the sexiest boring topics of this cycle. And AI can help tremendously to improve data quality. The data elements that you have in a normal ERP system, be it SAP or Oracle, it doesn’t matter – you have a lot of transactional data.

Every single day you have transactions that you can use for process intelligence, and you can also use them to learn about the parametrization of the supply chain. A simple example is lead time. So, if I order a part from a supplier, there is a lead time until it arrives and that is the time-stamp that I track.

And it's very easy to start analyzing those lead times from a supplier and comparing that to the master data. "What does the master data say? What is the ‘should’ and what is the reality?" And I start to optimize, to analyze, and to then predict for a given supplier how they're performing at the moment, whether there are trends. Because this shouldn’t be static anymore – it should be predictive and dynamic.

Q: Speaking about supply chain and growth, what are the 2026 priorities? Where do you see opportunities?

Julian: One of the priorities is definitely resilience. There has been such a tremendous amount of volatility over the last few years. It started, and really became predominant, with COVID. And after COVID, I remember a good friend of mine, who's a chief supply chain officer, asking me, "Will we go back to normal?" Back then I remember saying, "I think so." Now I know I was wrong. The amount of uncertainty that there is in global supply chains is huge. Take the tariff situation that we have at the moment. Many organizations do not really understand what it means for them. So resilience will remain key.

The second priority is using agentic AI. Do I have teams that are composed of normal ordinary planners, that are augmented by agents for specific tasks and activities? This will disrupt the way organizations work, enable professionals to be more focused on value-adding tasks, and change the capability landscape of these organizations.

Q: How can supply chain leaders communicate risk to the boardroom?

Julian: That’s one of the hardest challenges. You can’t build a traditional business case for risk the way you can for, say, a new factory or efficiency project. You can’t easily put probabilities and ROI on black swan events.

But what we do see is that investing in resilience pays off. Companies that prioritize risk management are more likely to withstand disruption. Even if it’s not a neat “green line” on a business case, resilience has to stay at the top of the boardroom agenda.

author

• Omar Sharif Valdez

McKinsey and Celonis: find out more

To learn more about how McKinsey and Celonis are helping companies build resilient supply chains, check out the video interview series, download the Supply Chain Optimization Strategies e-book, or learn more about the McKinsey x Celonis partnership.