Goodbye Spreadsheets…Hello Predictive Analytics!
In today’s supply chain, disruptions, geopolitical issues, and adverse weather events are becoming more and more common. But the good news is – there is relief in sight! Companies are working to exploit the tremendous amount of data from their transactional systems, and combine it with data from other supply chain intelligence systems and real-time news feeds to get directional indicators of which areas of their supply chain are most vulnerable to reduced performance.
A single large company’s supply chain can be incredibly complex, spanning thousands of suppliers and tens of thousands of parts moving across more than 100 countries worldwide. Leveraging advanced knowledge and smart technologies to filter data into a concise area of focus are a game changer.
Speaking of predictions, I predict that, eventually, machines will automatically generate orders up and down the supply chain. They will do so based on an ongoing stream of new events, adjusting and accommodating to these shifts and notifying buyers and sellers on how to adjust production and shipment schedules accordingly. Planners’ reliance on spreadsheets will disappear, and instead planners will get a stream of options and questions that they can answer based on available data, and provide input into the system.
One of the key technologies that will enable this to occur will be the development of digital twin models. A digital twin is essentially a model of an existing supply chain system – which is updated in real time with actual data pulled from various IT systems across the entire network. The scale of digital twins has not yet evolved to their full potential, but we will likely see the day when everyone in the system will be able to see the current state of material movement across multiple tiers of buyers, suppliers, and distributors. Imagine if a supplier two tiers downstream, could see how many units of their product was selling through POS systems at major retailers – and be able to adjust production levels accordingly. They would be able to anticipate changes in orders that would be forthcoming, and machines would alert them of these changes, allowing rapid shifts up or down in capacity and schedule planning. What if they were also alerted to promotions occurring at the retail level, and when these would occur, and the likely probability of escalated volumes, in real time? And what if retailers could observe disruptions such as weather-related disturbances occurring upstream at a supplier of critical material, and understand the effects of this issue on supply of final products? And anticipate mitigation actions to avoid stock-outs, through alternative supply or working with other distribution channels? These types of scenarios will soon be possible, allowing real-time visibility, modeling capabilities, and probabilistic scenarios for risk mitigation planning using digital twins.
Convoy is a technology that works with truckers carrying freight. More than 80% of trucks on the road today are independent – and most of them are operating half full. This is because they travel on one leg of a route to deliver a shipment, and often return back the other way with an empty truck.
Technologies such as Convoy provide drivers with a menu of pick-up options to choose from, based on where they are located (much like Uber does with drivers). This automated feature will allow truck drivers to pick loads that are the most profitable and that suit their scheduling preferences, optimising the flow of materials in the supply chain based on updated price quotes, capacity and scheduling of vehicles.
As our planning systems collect more information about real-time events, updates to the system will become more frequent, allowing continuous adjustment in supply chain responses. Humans will still be very much “in the loop”, but a new set of analytical skills and decision-making based on incoming data will be required.
Another application of digital twins is to simulate different future-state supply chain configurations, and enable improved strategic business decisions. For instance, one of the biggest problems companies are facing today is the question of strategically aligning their global supply chains, to consider the cost of on-shoring, near-shoring, or continued off-shoring. This challenge remains challenging, due to the problems of quantifying the different types of costs associated with this approach.
A team of NC State students partnered with an on-demand apparel company to construct a new type of off-shoring vs. near-shoring cost model, which accounts for multiple costs that are often overlooked or difficult to quantify. This includes the cost of supply chain risks, cost of demand volatility, cost of being “wrong”, and inventory carrying costs. These costs were developed through in-depth discussions with apparel management executives, and the assumptions were validated and tested.
The model was developed for a single “SKU” – a non-common color sweatshirt. The results were astounding and provided real insight into the revenue improvements associated with a mixed offshoring-reshoring strategy.
The next steps for the model would involve further refining assumptions, expanding the model to include sustainability and technology impacts, and the additional of hundreds of SKUS and different sourcing options. In short, this model could benefit from the development of a “digital twin” model –which could be refined and tweaked to suit any type of retail consumer product category. Imagine if a retailer could mitigate shortage risks and model a more responsive and agile supply chain to maximize profitability and sales revenue? This capability will soon be available.
The new level of automation and smart sensing in supply chains will allow flows to become more predictable. Smoother flows will be the antidote to an increasingly bumpy, or uncertain, world.
