I am talking about going beyond using traditional historical data on past sales and stockouts. It is now possible to link data generated by all product interactions (including orders, examinations, and reviews by actual and potential customers) and transactions generated by suppliers and competitors who connect via internet web sites and cloud portals. This data can be used by material-management systems to control ordering and distribution of products throughout a company’s extended supply chain. In addition, any data that is coincident with these product interactions, that is derived from the firm’s external environment, can also be accessed and linked.
How will this work? Advanced machine learning and optimization algorithms can look for and exploit observed patterns, correlations, and relationships among data elements and supply chain decisions – e.g., when to order a widget, how many widgets to order, where to put them, and so on. Such algorithms can be trained and tested using past data. They then can be implemented and evaluated for performance robustness based on actual realizations of customer demands. For example, does use of these data-driven tools lower cost and/or enhance customer service?
Why does this matter? The traditional paradigm for supply-chain management is to develop sophisticated tools to generate forecasts that accurately predict the value and the level of uncertainty of future demand. These forecasts are then used as an input to an optimization problem that evaluates trade-offs and respects constraints in order to come up with decisions about managing materials. This two-step process, which is embodied in all current material-management planning and control systems, can be replaced by a single-step process that looks for the best relationship among all of the data and the decisions. Based on learning from the past, a “best” relationship can be identified, which will generate decisions, as future uncertainty is resolved, that are better than the decisions derived from the traditional two-step approach of first forecast and then optimize.