Given the global supply bottlenecks of recent years, supply chain risk management is increasingly gaining strategic significance. Irmgard Sturm, department lead Process & Data Analytics at msg describes the role artificial intelligence plays in identifying early warning indicators and why speed is not the only factor.
Over the past three years, we have experienced various crises that had a massive impact on global supply chains. Couldn’t these events have been foreseen? Why were companies seemingly so ill-prepared for them?
Irmgard Sturm: Disruptions in supply chains are not unusual in themselves; they are a recurring problem. Usually, consumers hardly notice this. Especially since companies whose supply chains have already been exposed to major risks in the past often have active supply chain risk management in place to cushion the impact of such external shocks. What we saw during the pandemic is that our globalized supply chains were affected simply by the sheer volume of simultaneous disruptions all over the world, to the point that it impacted almost every consumer. Could this have been foreseen? In all its complexity and with all its implications, certainly not. But a crucial question is, how do we recognize for our company-relevant segment the likelihood of a particular risk occurring, the potential impact , and what should we be willing to invest to avoid the risk?
So, what should companies do?
Irmgard Sturm: In the past, companies have heavily optimized their supply chain towards efficiency by minimizing their storage or by consolidating their supply relationships, for example. This was done for reasons of efficiency and because the probability of these global risks was considered low. Many companies are now trying to move back in the direction of resilience, for example by establishing or expanding a multi-supplier strategy or by bringing parts of the value chain back in-house or to Europe. These are reactions that acutely benefit companies, but of course have been decided differently in the past for good reasons. In this respect, it is difficult to switch from the previous efficiency strategy to a resilience strategy in the short term. All in all, each of these strategies should be viewed critically on its own merits. A company that is optimally positioned in terms of efficiency will have a problem with every disruption. And those who are optimally resilient are unlikely to be competitive in the long term.
What could be a way out of this dilemma?
Irmgard Sturm: The idea is to identify the relevant risks that can actually influence one's own business and to find early indicators for them in order to be able to react, if possible, one day - one week - one month earlier than the competition and with as many courses of action as possible. This should enable companies to develop measures proactively and not to react only when the risk has already occurred. In addition, it should be possible for companies to decide in advance what they are willing to invest in order to be prepared for these risks. This can be evaluated using historical data, for example: What triggered the supply chain shock in the past, what were the indicators, and what was the impact of the shock? Systematically investigating these questions and making them a recurring process is the idea behind our “Supply Chain Shock Prevention”. Preparing for resilience and ensuring efficiency - that's the goal.
How does this shock prevention work in practice?
Irmgard Sturm: Shock prevention describes a method toolset whose elements can be used individually or interlocked. In essence, the aim is to systematically identify threats at an early stage, if possible with the help of AI, assess their impact and weigh up or prepare alternative courses of action. The idea of the method toolset is to understand supply chain risk management individually, but systematically as a recurring process. For this, the individual method elements interlock in a logical sequence. This begins with requirement forecasting, continues with the identification, evaluation and analysis of shocks, the simulation of possible consequence and optimization options, and ends with the creation of a prevention plan. The example of the threat analysis of a commodity, e.g., cocoa, can be used to illustrate this very well, albeit in a highly simplified way. On the one hand cocoa, as a natural product, is exposed to strong environmental influences and thus to the risk of crop failures; on the other hand, it is only grown in a few regions, which increases the risk of influences such as strikes, political or social unrest, etc. Both risks that can potentially impact availability or prices. To be able to perform preventive risk management here, one would first investigate whether there have been recurring events in recent years, which have impacted prices. These events can be categorized, for example, by political, environmental or economic influencing factors. For each of these categories, AI can then be used to simulate how future events are likely to affect prices using historical data. This, in turn, enables companies to define early warning indicators and threshold values at which certain measures must be taken.
To what extent has the use of artificial intelligence changed the possibilities of such simulations?
Irmgard Sturm: AI makes it possible to analyze large volumes of data objectively, in a structured and automated way. This not only makes people's work easier, but also increases quality through objectivity and makes it comprehensible through standardized procedures. This way, AI has massively extended the analysis, forecasting and simulation possibilities by allowing more and more data to flow into risk analysis and be processed efficiently - and thus making new correlations visible. In principle, everything for which there is a data basis can be included in such forecasts or simulations. The difficulty here is to find valid data for the respective scenario. The transparent supply chain is still a dream - in purely data-related terms. It is therefore necessary to look at the supply chain successively from the company's point of view and see what data is actually available, what external data sources can be included and how this data can be meaningfully linked together.
Even if access to data is sometimes restricted. The number of possible questions that can be addressed in the context of such simulations is almost unlimited. Where would be a good place to start?
Irmgard Sturm: This is actually a very good question, which every company has to answer for itself. The supply chain shock prevention method offers one approach here: demand forecasting. If not already available, AI can be used to identify future demand based on company-internal data in combination with market data. After all, shock prevention must look into the future and thus must know the relevant future needs. Perhaps at this point a few words about the methodology itself: Supply chain shock prevention provides a method toolset with interlocking analysis steps based on proven and innovative AI methods. It is thus, on the one hand, a guideline for the structured analysis of potential supply chain shocks and, on the other hand, as a result, is an automated, AI-assisted toolset for monitoring supply chain risks.
Let's take the example of the automotive industry where the lack of electronic components was a huge issue during the pandemic. But there, too, the question now arises: If I want to switch to e-mobility in the future, what are my critical elements? Maybe it's still the chips, maybe it's the batteries or components of batteries that are a bottleneck. And I'll probably identify a lot more than one critical element in my supply chain. Maybe it's thousands of critical elements in the automotive industry. No matter how competent and well-equipped organizational units are, this can no longer be monitored without innovative methods and a structured approach.
Is it even possible to set up something like this from within the individual company in a meaningful way, or does shock prevention not necessarily have to be thought of in ecosystem terms?
Irmgard Sturm: Ecosystems are once again a separate issue. Efforts are diverse and in some cases promising. Considering that after the direct supplier (Tier 1) the transparency of the supply chain for companies often decreases significantly, cooperations beyond this are essential. There are already efforts by platforms to address this very issue in order to create more transparency and achieve higher data quality. The more usable data is available and the higher the data quality, the more meaningful are the possibilities for shock prevention.
But if everyone benefits equally, isn't that a zero-sum game? In the end, isn't it about being faster than the competition?
Irmgard Sturm: Of course, it's also about speed. First, the procedure can be used to give the customer a competitive advantage. Even if, at some point in the future, all companies in the world or an industry use our shock prevention, the courses of action found will continue to be diverse and have a coordinating character. Above all, it's about having courses of action. We have a competitive situation, of course, but we also have several courses of action, some of which can be well prepared. And not every company acts the same way. Whereas company A decides: “'I can afford to stock more, so I'm prepared for this shortage,” company B might say: “Warehousing is not an option for me because it's just not attractive price-wise anymore.' I would rather think about a multi-vendor strategy. However, when resources are scarce, competitive situations cannot be completely ruled out. If two companies want the same product, the luckier one, of course, is the one that was faster. But the probability that all courses of action found will lead to the same result, as in the highway example, is rather low. The matter is too complex for that.
Even if not all market participants behave in exactly the same way – doesn't shock prevention always have to take into account the behavior of competitors?
Irmgard Sturm: In narrow sectors, of course, this can become relevant. Let's take the example of cocoa again: Cacao is only grown in a few regions. If, for example, the Ivory Coast can no longer deliver, all chocolate manufacturers would have a problem. If I know this one month in advance and can replenish my stock, I have a competitive advantage. Or I have estimated this risk to be high, but don't want to or can't build up stocks, in which case it's good to have thought of alternatives. To stick with this simple example, we are now producing more white chocolate and already have a marketing strategy in place for this. This means that shock prevention is actually more of an option initiative as a mere race. Put another way: If all companies prepare for pandemic 2.0, then everyone will be doing the same thing. But that's not the idea. Rather, the idea is to make organizations and their supply chains more creative, agile, and company-specific between efficiency and resilience.
Interview partner
"It' difficult for companies to switch from their previous efficiency strategy to a resilience strategy in the short term."
"AI has massively expanded analytics, forecasting and simulation capabilities by allowing more and more data to be incorporated into risk analysis and processed efficiently."
"Above all, it's about having courses of action."