The use of machine learning in supply chain management has been the subject of much hype. Has its time finally arrived?
A recent Wall Street Journal article profiled The Absolut Company, makers of Absolut Vodka, the world’s largest locally produced vodka, manufactured in the small town of Åhus, Sweden. The operation produces 11.5 million nine-liter cases of vodka each year, with 75% of production shipping from the small harbor of Åhus. The company keeps a tight production schedule and follows lean production principles, receiving deliveries every hour and maintaining just three hours of safety stock on hand. Absolut faces an increasingly complex production environment: over a five-year period, the number of SKUs rose by 19%; the number of core flavors increased from 11 to 18; and the number of limited editions went from two to 12.
The foremost of these supply chain challenges was the increased production complication. Additionally, in contrast to its highly automated production process, the production planning process was a manual task, with forecasting and production planning being performed by a single planner with the help of spreadsheets. There was also a need to better support the company’s lean production process, and a desire to reduce inventories despite growing sales. These are all familiar issues in the global electronics supply chain as well.
The company looked for a better, more efficient way to manage production. It implemented a fully integrated production planning solution, which employed a high level of automation and utilizes machine learning to optimize its production and supply chain management. The new forecasting and production planning process still requires only one planner and has resulted in the ability to cut inventory by 20%.
The Gartner Hype Cycle published in July 2016 showed machine learning approaching the Peak of Inflated Expectations, and predicted that mainstream adoption of machine learning was at least five years away, potentially 10. Yet, success stories such as The Absolut Company show that machine learning can be an effective technology to improve supply chain management performance today. Its use can help revolutionize the optimization and agility of supply chain decision making, yielding benefits such as the abilities to:
- Improve production planning and factory scheduling accuracy by considering multiple constraints and optimizing for each
- Reduce freight costs, improve supplier delivery performance and minimize supplier risk
- Lower inventory and operations costs
- More quickly respond customers
- Improve supplier quality management and compliance by finding unassisted patterns in each suppliers’ quality levels
What exactly is machine learning? It’s a type of artificial intelligence that gives computers the ability to learn without being explicitly programmed. Machine learning computer programs teach themselves to grow and change when exposed to new data. By doing so they can discover new patterns in supply chain data without manual intervention.
A good place to start using machine learning is in demand planning by employing “best-fit” forecasting, a basic form of machine learning. A best-fit forecasting algorithm automatically switches to the most appropriate forecasting method based on the latest demand information, ensuring an organization creates the best forecast for every product at every stage of its life cycle. Another good place to start is the use of software solutions that use algorithms to continually analyze the state of a supply chain and recommends or automatically executes plans to meet customer requirements.
Perhaps it’s time for your supply chain organization to explore the use of machine learning. Implementation of machine learning is an evolutionary process, with its full benefit being realized over a period of years. Yet, by taking advantage of machine learning tools that are current on the market you can build expertise and experience that will ease your adoption of advanced solutions in the future.