Weathering The Storm: Risk Intelligence Edition

Scott Pezza

In recent years, we have seen numerous extreme weather events across the globe that have had significant impacts on both individuals and businesses. From hurricanes and tsunamis to earthquakes and volcanic eruptions, these events often come with little warning and cause serious and long-lasting logistical issues. In this blog, we’ll look at the challenges these events pose and highlight how technologies like artificial intelligence can help mitigate – or eliminate – their negative impacts on your business.

Recognizing the event

It may seem simplistic to begin a discussion of “extreme” events by talking about recognition. By definition, they are large-scale and obvious. The most important element in this area, however, is focused on when you recognize the event. Here’s a helpful breakdown that lays our foundation for further discussion based on three alternatives. You can recognize that the event:

  • Has occurred. This is purely backward-looking. This is how we normally recognize events, especially when they do not impact on immediate geography. As individuals, we gather this information manually from newspapers, websites, and social media based on the first-hand observations of others.
  • Will occur. This is forward-looking, though typically limited to a few days or hours before the event. Aside from the most extreme or large-scale events, this typically comes in the form of warnings and alerts from governmental organizations. We may get this information from weather services, local governments, local news services, or official social media channels.
  • May occur. Here, this forward-looking view is based on the conditions that are known to have resulted in extreme events in the past. For some types of events (for example, weather-related), these look similar to the category above and come from sources like publicly available services. For others (for example, geopolitical events), any advance warnings are more likely to come from non-public or subscription-based third-party data services.

Understanding the impact

At this point, we have recognized that something has, will, or may occur. The next step is to understand what that means from a practical point of view. Weather events like hurricanes and floods will disrupt transportation routes and threaten electrical infrastructure. The ash from volcanic eruptions, as we’ve seen, can significantly impact airline routes far beyond the local eruption. Geopolitical events can impact these same areas while also introducing the possibility of a political or policy-based prohibition on certain activities in a geographic region.

As before, we can break down our impact analysis into three categories:

  • Immediate impact. This is the starting point. Whether observing after the fact or predicting ahead of time, we need to know what the impact is on Day 1.
  • Potential to expand. This answers the question of “what next?” Forest fires can spread great distances from their point of origin, and storm systems can travel hundreds or thousands of miles after making initial landfall. This applies predictive models to current observations to plot out potential expansion.
  • Challenge to address. Based on the scale and severity of the impact, we need to understand what is required to regain operational stability and how long that process may take. Localized power disruptions can be remedied in hours or days, while repairing damage to infrastructure like roads and bridges will likely be measured in months or even years.

Making crucial connections

One final element to cover before bringing all of this intelligence together is determining which of your locations, suppliers, sub-suppliers, customers, or trade routes are – or will be – impacted. This is where the complex nature of global supply chains and logistics networks can make things difficult to ascertain. The easiest group to account for is the list of locations for your own business and subsidiaries. Direct suppliers come next, though with additional difficulty when your vendor master lists its business (or billing) locations rather than lesser-known locations like manufacturing or excavation sites. Moving beyond direct suppliers to their suppliers, and their suppliers’ suppliers, exponentially compounds the challenge.

Disruptions can come at any point along the chain. Your retail outlet may suffer stockouts if a distributor’s warehouse is affected. Your production line may meet a similar fate if your raw material or subassembly supplier is hit with delays. Without a comprehensive supply chain mapping complete with geolocation data, anything beyond an analysis of direct and tier 1 suppliers is likely impossible without the aid of some software tool – and that’s assuming that you have perfect vendor master information to work from.

Synthesizing information for action

With that background, let’s see how AI tools can help bring everything together to help mitigate the impact of an extreme event. Imagine that a strong weather system is developing off the east coast of the United States, 500 miles east of Florida. You operate 50 retail locations across the Northeast and Great Lakes regions of the U.S. What do you do?

  • You have enriched your vendor master to include not only billing but “ship from” locations for all suppliers. You’ve leveraged a third-party service to map all suppliers using a “global ultimate parent,” creating a hierarchy of related companies.
  • A weather service data feed sends text alerts with updates on the projected storm path. Using natural language processing (NLP), your system converts the alerts into data on projected locations and weather conditions.
  • Using a trained machine learning (ML) system, you project the expected severity of impact at each geographic location and predict how businesses in that region will be affected.
  • This process creates an impact map of geographic locations and expected disruptions. This is used as an overlay to your comprehensive supplier map to isolate and identify which suppliers are in the storm’s path.
  • You see that the storm is expected to turn north before making landfall, with a low likelihood of impacting your inbound shipments to the Port of Savannah. Your imports are safe.
  • The system sends an alert, however. Your Great Lakes regional distribution center is close to reaching an automated reorder point for an important piece of merchandise. That product is normally sourced from a distributor in Maryland that receives its imports from the Port of Baltimore – which is directly in the projected storm path.
  • Your supply chain collaboration software shows that this supplier’s existing inventory will not be sufficient to meet your upcoming order volume – and that the storm means that it is unlikely to receive enough inventory in time to fulfill your next order.
  • The system identifies a second supplier with adequate inventory to fulfill your order from California, leading to higher inbound transportation costs.
  • By integrating with your financial system, you know that each day of stockouts on this item will mean $50,000 in missed revenue. Based on the current storm projections, you expect at least a week of delays – or over a quarter of a million dollars in lost sales.
  • With detailed models of weather systems, supplier locations, inventory levels, and with integration to financial systems, you are confident in executing a replenishment order with the second supplier. An extra $25,000 in transportation cost will be well worth it to avoid 10 times that amount in lost revenue.

This is just one example of where new technologies like AI can help address problems that were incredibly difficult previously. There are many more use cases in risk intelligence and beyond, where cutting-edge solutions are redefining what is possible in the business context.

If you’re looking to learn more about what we’re doing in these areas, come to SAPPHIRE NOW in Orlando June 5-7. To give you a preview of just some of what we have in store, take a look at all of our procurement-focused sessions here at this link.

Scott Pezza

About Scott Pezza

As part of SAP Ariba's Digital Transformation Organization's Center of Excellence, Scott researches, compiles, and shares best-practice information to help SAP Ariba's customers get the most out of their investments. He has a dual focus on the emerging technologies (AI/ML, IoT, Blockchain, etc.) across the source-to-settle cycle, as well as a specific interest in the financial supply chain (invoice management, payments, discounting, and supply chain finance). His research helps inform strategic planning, performance measurement, and program execution. He has spent the past 17 years in the B2B technology space, in roles ranging from software development and support to research and consulting. Scott earned his BA in English and Philosophy from Clark University, his MBA from Boston University Graduate School of Management, and his JD from Boston University School of Law, where he served on the Executive Board of the Annual Review of Banking and Financial Law.