Author: Betsy Romeri
Your company’s historical data are not contextual enough for our COVID world.
Danillo Pereira, CAO
June 22, 2020
The fallout from the COVID-19 pandemic on businesses has made it obvious that global supply chains are more fragile than we realized. Businesses have traditionally relied on their own historical data of customer behavior to determine where they store their inventory and what level of demand they may have. But COVID-19 has taught us that analyzing historical data (yours and your competitors) will get you only part of the way to a prediction of the demands on your business during a surprise event. You will also need real-time external data that reflects what is happening in the world and AI models that predict optimal corrective actions.
Global newsfeeds that track political turmoil, social sentiment, developments surrounding natural disasters, live weather, and much more are game changers when it comes to predicting changes in global supply chains. We have also learned that social media and its “citizen journalists” also play an increasingly significant role in discovery of hot spots and impacts. Think of the videos and photos that citizens have been “on the spot” to take and their impact on the general public’s awareness of issues once those images are disseminated by mass media.
Today, there are numerous data companies that provide real-time data feeds of global news and events. Investors, businesses, and governments use these feeds to get early warnings of things to come. In fact, it’s no longer uncommon to see city emergency-service companies reporting that they learned of emergency situations via their Twitter feed before 911 calls came in. There are AI models that can filter through millions of Twitter feeds and pull out the relevant information for contextual intelligence for a specific business.
With early warnings of events from these and other sources, supply chain managers can take action before a problem becomes endemic. The technology to avoid stock-outs, overstocking, unfavorable contractual lock-ins, and poor pricing strategies exists today. But most businesses that have successfully crossed over to data-driven operations still struggle to forecast demand accurately.
Those who work every day in the supply chain arena also know that there are many supply chain decisions made using “gut” feelings or untested assumptions. The domino effects of such decisions can be immense. By using sensors of external conditions and events in your markets in the form of data streams, supply chain managers have an opportunity to realign their dominos.
AI models will always be more accurate than gut feelings from scant data because they are objective and have “considered” a wide variety and volume of relevant data. Natural language processing allows data scientists to take newsfeeds, social media text, and publicly available business reports and convert them into data for decision making. By training AI models to learn patterns in all sorts of contextual data that pertain to a particular industry, supply chain managers are able to identify the events that will lead to disruption. They are also able to prescribe the best actions to take to avert disaster in their supply chain or pricing structures. Some AI/ML models can automate an appropriate response or alert the human in charge to make the necessary decisions to avert disruption. These might involve stockpiling inventory, moving inventory, altering prices, or other actions. They might initiate alerts to all suppliers to coordinate a response to the potential threat to the end-to-end supply chain.
COVID-19 has made it plain: Supply chain managers would be wise to build external datasets into their AI models if they have not done so already. With continuous interaction with data drawn from the broader environment, AI models can predict behaviors that might endanger your supply chain operations, increasing your opportunity for corrective action.
Danillo Pereira is Analytics2Go, Chief Analytics Officer. In addition to multiple degrees in the computer sciences, he has published widely in the field of analytics, machine learning, and AI. His main focus at A2Go is creating modern analytics models that aid business planning and decision making.