Consider how fast China has adapted ecommerce platforms to the tune of $1.7 trillion in 2020. This is a non-trivial number reflecting 30 percent of all retail sales in the country. Step back and examine how Amazon, Walmart and the rest of the ecommerce platforms are impacting the supply chains. They are putting tremendous amount of pressure on the producers and manufacturers to deliver or die. They have already set consumer expectations for “perfect order” delivery, i.e., right product, to the right customer and right place at the right time and quantity. The right time currently seems to be next day or even today! In a couple of years, it could be “now.”
Companies talk about being responsive in their supply chain. However, it is not sure if they have time to do much when they have only hours to respond and seconds to commit. Agility has many different implications in design and operation of the supply chain. Agility means having a resilient supply chain to be able to fulfil orders despite all the disruptions that we have been facing, and keep growing. Agility also means being faster than your competition to make and deliver. It also means quickness to adapt as the environment changes, consumer demand patterns fluctuate and competitive landscape gets tougher.
To deal with the exciting times ahead of us, just designing a supply chain that can respond is not good enough. It needs to have the ability to predict, not just respond. Predict potential demand fluctuations, predict consumer behavior, predict supplier conduct and performance variations, predict weather patterns, and predict new product trends and so on. Quality data, not just any data is the first step. Deploying such data and using AI/ML techniques, we are at a point to detect the pattern changes in structural and behavioral changes in the supply chain and figure out what the causes are. Thus, mitigating the impact of disruptions and improving customer service. If a supplier promises delivery in 2 weeks; it is possible that after a few months or perhaps years, their expected delivery is changed to 10 days. This could have enormous implications in reducing inventory and delivering sooner to the end customers. By the same token, we can estimate and predict availability of resources and plan more realistically to generate the right amount of inventory and reliable commit dates.
By understanding multiple causes for demand changes, we can now readily predict what consumers want, where and when and have them ready even before they know what their needs are. Safety stocks are subject to many different factors. Amongst them are equipment and resource availability, seasonality, and competing products. Machine learning algorithms have the capability to predict almost exactly what the right quantity of each SKU, intermediate or raw material should be where and when.
Not everything is predictable. Pandemic taught us that lesson. A major container getting stuck in the Suez Canal is another unpredictable one. Compare these events that are high cost and low frequency to the ones which are low cost and high frequency. The latter is costing companies hundreds of billions of dollars every year. Much of it can be saved by better prediction and design of the supply chain to make them more resilient. That is avoiding dependency on vulnerable suppliers and regions. Keeping the right amount of inventory in the right places as a result of better prediction and having systems that can “re-program” the supply chain. Thus, the digital running the physical.
According to the CEO of a freight company “No matter how digital you are, cargo has to move physically from point A to point B.” This statement assumes our dependency on things that we may not have control over. Thus, allowing the physical to drive the digital! This is old school. Perhaps the “digital” could find the alternative way to deliver to point B with much better efficiency, lower cost and lesser risk!
To learn more about application of AI and ML in supply chain planning click Here.