Since 2011, I’ve been teaching Supply Chain Risk Management at Lehigh University. Developing a new MBA class in Supply Chain Risk for our supply chain MBA students gave us the opportunity to explore the landscape of Supply Chain disruptions and the statistical tools and techniques in the areas of Predictive Analytics and Big Data that are improving supply chain performance. Let’s start with the data landscape as of today, and chat about a few use cases where Big Data has real, practical applications.
According to IBM’s 2014 Big Data Study:
- More than 90% of all data that exists in the world today was created just 2 years ago!
- We’re in an age where more than 2.5 quintillion bytes of data are created EVERY DAY!
- That’s more data than all the data in all the libraries in the USA
Here’s IBF’s definition of Predictive Analytics:
“A group of statistical techniques including modelling, machine learning, and data mining that are used to analyze current and historical data in an effort to create projections about the future.”
The exciting aspect of Predictive Analytics in supply chain and risk management is that the computing power has now caught up to the algorithmic strength in the discipline, creating huge opportunity to leverage these age-old tools to enhance supply chain performance and mitigate supply chain risk.
Use Cases Reveal Predictive Analytics To Be a Growth Driver
As of today, there are over 20 different industries exploring the use of Predictive Analytics and Big Data. A recent E&Y study stated that 66% of companies with well-established advanced analytics are reporting operating margins and revenue growth of 15% or more. And 60% of the respondents said they also have improved their risk profiles. Very exciting! Let’s take a look at some quick-hit use cases.
Starting in 2016, Lowe’s began to utilize autonomous retail robots to help both customers and store staff. It helps customers find products while it helps staff by scanning items and identifying price discrepancies. More interestingly for us, however, is that it tracks inventory. It knows what is being sold and where, all in real-time. That provides great visibility into demand and supply, enabling its supply chain to react faster. The robot is bilingual and driven by Predictive Analytics – and could soon be replicated in a variety of retail environments.
IBM and Watson
Accenture’s 2016 report into the Internet of Things concluded that Predictive Analytics could save up to 12% on scheduled repairs and 30% on maintenance, while reducing breakdowns by up to 70%. When it comes to reducing breakdowns, this is where Predictive Analytics is truly remarkable. See The IBM commercial below, where Watson predicts that an elevator will malfunction in 2 days, allowing the repairman to ‘fix’ it before it breaks.
IBM, a big advocate of end-to-end supply chain management, Predictive Analytics, Big Data and SCRM, Supply Chain Risk Management, is leveraging Watson to enhance their SCRM approach. IBM uses Watson to identify, assess and mitigate supply chain risks, 24/7, 365 days a year.
IBM, The Weather And Retail
Consumer behaviour is strongly influenced by the weather. IBM, who bought The Weather Company, is using Watson to make stunning predictions about the weather. Now, making predictions about the weather is nothing new but using integrating the findings for better supply chain decisions is. IBM uses databases from The Weather Company and a variety of of Big Data, including news stories and social media to figure out in real-time how consumers are reacting to weather conditions, and the subsequent commercial opportunities.
Their weather predictions coupled with pattern recognition tools are providing forecasts months in advance, which are then integrated with retail companies’ supply chain data. Armed with this information, companies can arrange resources accordingly, with Demand Planners able to make more accurate demand plans, and Sales and Marketing able to promote certain items and price accordingly (Figure 1).
Jon Walker from Tech Emergence reveals that Walmart has seen some very interesting (and highly profitable) correlations between weather and consumer spending behavior. The company noticed that people tend to buy steaks when it’s warm, windy and cloudy. Why exactly? We don’t know, but they do. Meanwhile, hamburger sales increase in hot and dry weather. They utilized these correlations to promote hamburgers based on weather predictions and saw an 18% improvement in beef patty sales. Go figure!
Anecdotally, we know (broadly) how weather will impact spending behavior for different industries, but never before have we have had this volume of data to support findings, this ability to reveal cause and effect, this ability to remove bias, nor this ability to automate the process and integrate into supply chain systems.