Analytics in supply chain management may be critical to mitigating inventory shortages and evaluating customer demand as coronavirus continues its global spread.
From shortages of personal protective equipment to a variety of grocery items to electronics and apparel, coronavirus (COVID-19) has hit the global supply chain in expected and unforeseen ways, and it seems likely that it could take many months to recover.
Bouncing back more quickly, said experts, will require supply chain managers to turn to new ways of managing the supply chain, including using Internet of Things (IoT) data, analytics and machine learning (ML). These tools will become the foundation on which supply chain managers gain insight into their markets and erratic supply and demand trends.
“Having the right machine learning and AI technologies will help you understand the market and better manage your supply chain,” said George Bailey, director of the Digital Supply Chain Institute.
While the disruption is now global, its starting point was in China — the 800-pound gorilla in global production. Indeed, by 2010, China surpassed the U.S. in manufacturing dominance. And while during the SARS epidemic of 2002 and 2003, China represented 4.3% of worldwide gross domestic product (GDP), today, said MIT professor David Simchi-Levi, the country represents 16%.
‘Globalization As We Have Known It … Is Over’
Manufacturing companies that have relied on China for production materials are feeling the blowback of this dependence; some retailers source more than half their inventory from China, according to 2020 Statista data. Another Statista study indicated that 44% of retailers expect delays and 40% expect inventory shortages given coronavirus disruptions on the supply chain. And more than half of electronics manufacturers anticipated up to four weeks in supply chain delays. That’s a difficult pill to swallow in an era when customers expect two-day delivery.
Now, companies are scrambling to assess their supply chain, but in truth, managing risk in the supply chain hasn’t been companies’ focus. The Institute for Supply Management, which conducts monthly economic surveys, found that nearly three-quarters of the companies it contacted in late February and early March reported some kind of supply chain disruption. But 44% of respondents didn’t have a plan to deal with it.
“What’s changed now is the sense of urgency to diversify, to have redundancies,” said Alex Capri, a visiting senior fellow at the National University of Singapore’s business school, in a CNBC interview on the importance of localizing value chains. “Globalization as we have known it in the past is over,” he said.
Bringing Analytics to Supply Chain Management
What is vexing for many manufacturers is that not just their supplier but also their suppliers’ suppliers’ have experienced plant closures, inventory shortages, transportation delays, worker absenteeism and so forth. So manufacturers’ tier 1, tier 2 and tier 3 suppliers are all experiencing disruption with reverberating effects throughout the supply chain. But they don’t necessarily have visibility into all those disruptions and how they will affect their own supply chian.
“Since almost all manufacturing companies have a substantial portion of their supply chain based in China, either directly or through tier 1 or tier 2 or tier 3 suppliers, capacity has dried up because factories are closed or understaffed,” Bailey said..
According to a 2018 Statista survey, visibility into that chain is a significant organizational challenge for 21% of supply chain professionals.
“Today most companies use Excel to put together different scenarios,” Bailey said. And “it’s a great tool,” but there are more sophisticated, more accurate tools to do sourcing. More than 90% of supply chain managers, use Excel somewhat to heavily for supply chain analytics. About 82% that use advanced analytical tools, according to Supply Chain Quarterly
While Bailey puts stock in AI as a future promise, designing systems to mimic human intelligence isn’t possible without sufficient data. An AI system needs to be fed data sets to learn how to behave and react. One-off situations pose a challenge, in that the system doesn’t have enough data to learn how to react. “In order to build a correct demand plan, one-off events have to be identified and accounted for,” wrote Ralf W. Seifert and Richard Markoff in the article “Demand for AI in Demand Planning.”
The authors also noted that successful AI in supply chain management is predicated on departments having consistent forecasts. Sales and operations, they argue, must operate from a single source of truth; otherwise, AI algorithms are prone to bias and inconsistency at the outset.
AI in Supply Chain Management: Demand Planning
Another key factor, Bailey said, is that demand planners need to better understand demand in times of crisis and help shape it.
Bailey said that IoT-generated sensor data becomes all the more important to gauge demand and manage supply. Tire companies, for example, now use sensor data to monitor tire pressure and proactively alert customers about maintenance. As they monitor tires’ tread wear and calculate its end of life , they can also send information to manufacturers about inventory needs and alert customers to manage the purchase.
“That’s using technology and analytics to understand what’s driving demand and using AI to estimate future requirements,” Bailey said. While some companies will have to invest in technology and analytics applications and in new staff roles, “overall, labor costs will be lower,” he said, because technology can supplement the often-flawed work of today’s demand planners.
Research firm Gartner predicted that at least 50% of global companies would use AI-related transformational technologies in supply chain operations by 2023.
At the same time, many companies will have to spend time going back to the drawing board on data cleansing activities. “Many companies have a huge amount of data that isn’t trusted,” Bailey said. “It might not be in the right format, it may be inaccurate, have definitional issues, or it may be biased. A huge amount of time is spent making that data good enough to use,” he said.
And again, data quality is a huge concern. “The most striking challenge to applying AI to demand planning is in the availability and accuracy of data,” wrote Seifert and Markoff.
Finally, Bailey and others noted that the reality in the short to medium terms is that supply chain managers will have to use data analytics to manage in a time of scarcity and uncertainty.
Bailey noted that companies will have to use data to guide their decisions. They may need to offer four SKUs rather than 50, he said, and focus on higher-value customers. Data should drive those decisions, he emphasized.
“There will be some product rationalization and reduction in SKUs,” Bailey said. “If a company has only 100 of x item and demand for 500, they are going to go through a process to decide, in a fact-based way, [which customers] to prioritize.”
Sourcing decisions will also change, Bailey emphasized. “Most companies have decided that an overconcentration in China is not a good idea. Finding the right way to balance where you put things on the map — to balance out the risks and opportunities — has become super-important.”
Ultimately, experts say, the coronavirus will force supply chain management practices and use of technologies that companies have been forestalling but that are now central to supply chain success.
“For better or worse, we now have a crisis that’s going to force people to change how they run their supply chains,” Bailey said. “The good news about that is these were changes they were going to have to make in any event.”
