Predictive Analytics vs. Pandemics – Winning the Supply Chain Battle
Striving for no missed shipments. Day in, day out, it’s not a straightforward task. Western Digital’s supply chain teams always need to stay one step ahead. But when COVID-19 hit, with record-shattering flight cancelations, supply shortages, border backlogs, and ever-changing government regulations, the company took on a new plethora of challenges.
Many of Western Digital’s customers are enterprise and cloud companies that make up the world’s data infrastructure. The company’s products are part of the data foundation that our world relies upon amid our new remote work, learn and play reality. Yet even amid the most challenging and unstable global logistics situation, Western Digital was able to uphold its responsibilities of volume commitments.
How? By applying predictive data technologies in new and innovative ways.
Managing Risk with Data
It started over a year ago when a team of data analysts at Western Digital’s Procurement Center of Excellence designed a predictive model. This model was designed to identify potential risks and determine their impact on the company’s supply chain. Western Digital operations span several factories in Asia where natural disasters such as earthquakes or typhoons are a reality. A predictive risk model could help better navigate these threats.
“We wanted to be able to predict the impact of adverse weather and geological events to guide our actions in real-time,” explained Christopher Wolf, a Predictive Intelligence Engineer leading Emerging Technologies in Western Digital’s Procurement’s Center of Excellence. He continued, “what would we do if a typhoon is moving in? Is there a way to see the forecasted path of the storm and see which of our suppliers or hubs are in that path? And then, can we see what P.O.’s we have from those suppliers and mitigate all risk before the storm makes landfall?”
When COVID-19 started to spread, Christopher and a team of data engineers started to re-engineer the predictive risk model to predict the path of the pandemic. The looming lockdowns and other social distancing measures put the company’s supply chain at risk, and Western Digital to flex the flow of inventories ahead of such events.
Was there a Predictive Peak?
Western Digital’s procurement team began working. In a matter of days, a new model was created. They called this iteration “Predictive Peak.” It analyzed COVID trends such as the growth and recovery rates of cases and cross-calibrated it with geographical data such as total population, mobility data, and susceptibility by region. The data model could not only predict the peak of COVID-19 cases, but also determine a date when the company could expect COVID cases to start declining.
The algorithms proved to be very precise. Predictive Peak was able to predict shelter-in-place mandates, and the rise of a second wave, with a margin of error of +/- 2 days.
The predictive capability not only gave the company’s supply chain management teams a unique advantage, but also enabled Western Digital’s executive team to make a variety of business decisions regarding supplier qualifications, manufacturing, and inventory positioning strategy.
“Using Predictive Peak, we were able to predict lockdowns in Japan, Malaysia, China, and the Philippines a week in advance, and could make critical adjustments to the supply chain,” said Ben Tessone, Senior Vice President of Procurement.
The Digital Command Center
As the pandemic accelerated, Western Digital worked with external carriers and internal teams to activate mitigation plans, sort through the details of whatever new country, customs, or airline restrictions had been put in place, and their impact.
Christopher and his data engineering team continued iterating the model to adapt to the rapidly evolving circumstances, and the needs of broader teams. Sharing the Predictive Risk capabilities across functions allowed Western Digital’s Global Logistics’ Center of Excellence to use the core model and data set to enable Predictive Transport.
With few flights going out, cargo shipments started being transported on passenger airline seats. A creative, but costly alternative. The foundation of predictive data allowed the company to effectively negotiate air cargo pricing with strategic carriers, and ultimately enabled a shared cost avoidance for both parties.
Ultimately, Western Digital was able to save $53M over two quarters! Without this action, the impact to the company, its customers, and potentially much of the world’s data infrastructure, couldn’t have come at a worse time.
A Fair Share
Western Digital’s Predictive Risk was built on a strong foundation of connected systems and connected manufacturing. While these data-driven systems give the company’s supply chain unprecedented resiliency and visibility into operations, Western Digital knew that many of the suppliers it works with do not have such capabilities.
“It was important to us to not only help ourselves but to also extend our newfound prediction capability to our suppliers,” said Matthew Smith, Senior Director of Global Transportation & Trade Operations at Western Digital. He continued, “if a supplier was anticipated to be in the predicted impact zone, we could caution them so they could adjust. New lanes could be opened if needed, we could help them by qualifying alternate sites, assist suppliers to move material or place orders before the event of a lockdown. This shared approach gave us a significant competitive advantage.”
Embracing Flexibility, Championing Data
The unforeseen demands brought about by the pandemic challenged Western Digital to embrace flexibility, and leverage data like never before. Coming up with new ways to apply data engineering, probabilistic modeling, and math allowed Western Digital to build a more resilient supply chain and deliver the critical storage technologies that its customers, and the world, depend on.
For more questions, feel free to connect with Western Digital