Effective supply chain is characterized by only two factors; timely results and cost effectiveness. And it is by improving these two factors that your supply chain can truly become optimized. The focus is usually on planning, sourcing, inventory, production, and warehousing and distribution which is all valid, but supply chain optimization really comes down on one main thing: Demand forecasting.
Obviously, demand forecasting affects everything else. All sourcing and inventory planning is done based on the prospective demand, all production planning as well, all warehousing and transportation to different geographical locations will all depend on forecasted demand. Therefore, forecasting accuracy has major financial implications. How much, in terms of dollars, is covered in an earlier article here: Companies Are Saving Millions By Forecasting Better.
Therefore, optimizing the supply chain will eventually come down to demand forecast accuracy. And that’s exactly where the revolution within supply chain management is taking place.
For decades, textbooks on supply chain management have explained the importance of forecasting and taught time series analysis with exponential smoothing being the most commonly used method. In some cases these methods can certainly be useful, and sometimes they’re even better than not having any calculations at all. But in most cases, by far, those methods are too simple to forecast well.
A typical organization has many different products or services, some with much seasonality, some with no seasonality, some with regular buying patterns and some with complex patterns. Some products are highly dependent on the sales of other products and even multiple external factors, which makes it even harder to apply any traditional tools to forecast the demand. Traditional time series analysis is simply a bad tool for the problem at hand in most organizations.
With the use of artificial intelligence (AI) techniques, particularly deep learning (neural networks), forecasting accuracy is going through the roof compared to the aforementioned more traditional methods. By utilizing this sort of technology, companies are saving millions within the supply chain.
Much of the hidden cost is within sourcing/procurement. If the raw material arrives too early, then inventory cost is accounted for and the actual cost of it increases. If the raw material arrives too late, it can impact planning and increase the cost within the production and logistics, and in the worst case, lead to stockouts and lost sales.
Moreover, sourcing can have seasonal benefits, i.e. raw materials might be cheaper at some point allowing for optimal buying points, the same goes for exchange rate fluctuations if sourcing is international. Therefore, buying at the right time is critical as it can have major financial implications.
Again, organizations don’t want to have too much nor too little inventory. The inventory levels need to be optimized which can only be achieved by knowing how much is expected to be sold, i.e. using advanced demand forecasting techniques.
Once the demand forecast is accurate, organizations can use data concerning security stock, minimum order quantities (MOQ), and lead times. What’s even more impressive is that organizations at the forefront of using AI in supply chain management are forecasting the lead times as well, making the supply chain even more efficient and optimized. Research conducted by Marisa Brown, Director, Knowledge Center, APQC, reveals that the inventory carrying costs is normally between 7%–16% which shows that there’s a real opportunity at hand to reduce cost significantly.
The further ahead production can be planned, the better. A surprise order from a client will sometimes push the production plan to the side in order to deliver at the right time with overtimes and all the extra effort included. This significantly increases production costs and is typically not accounted for.
Furthermore, what’s not produced is not sold. Therefore, in order to avoid stockouts organizations need to have optimized production schedules where products are produced at the right time. And if it’s produced too early, it’ll accumulate carrying costs. Therefore, the right time is never too early and certainly not too late.
In production, surprises are most unwelcome. But now, as of 2020, organizations are adopting advanced forecasting techniques based on deep learning in order to optimize all production planning to the minute.
Warehousing & Logistics
Shipping your products to different geographical locations for warehousing is all about planning. Warehousing is expensive and shipping the wrong amount - too little or too much - will include higher warehousing- and transshipment costs for getting the products to the right location. Clearly, planning this depends solely on the accuracy of the demand forecasts.
Obviously, the supply chain is about the right-time deliveries and cost effectiveness, where the two are certainly intertwined. All the components of the supply chain discussed here above will always come down to less error and more accuracy in the demand forecast. The supply chain can never be truly optimized without it, this is where organizations should be placing their focus.
Never before has it been possible to reach the level of forecast accuracy we’re witnessing today with the use of AI, and business executives are taking notice. More and more companies are adopting these methods where the growth in this practice significantly surged throughout 2020 and the Covid19 pandemic. Obviously, business leaders are fighting to become more agile and optimized, saving money where possible, accelerate decision making, and become more competitive.
The pressure is mounting and the competitive landscape is certainly not becoming any easier. Organizations need to place their focus on becoming more cost efficient and the obvious place to start is to strive for supply chain optimization powered by sophisticated deep learning technologies.