Demand Forecasting for Inventory Management
Any company that carries inventory is forecasting, whether they think so or not. Many companies claim that they don’t forecast, but if you are stocking raw materials, work-in-process or finished goods in anticipation of fulfilling demand, you are forecasting. Often when companies are trying to decide how they can improve their supply chain strategy, the first thing they want to try is to develop an improved demand forecast. Due to seasonality, new products, lead time, or changing customer demand, simply ordering what was sold in the recent past will not work in most cases.
At Supply Velocity, as supply chain consultants, we understand demand forecasting and ultimately how it allows our customers to make improved inventory decisions. We begin by working with businesses to understand the underlying data. One distribution center (DC) serving 12 branches needs a different demand forecast than trying to figure out the right inventory levels for a single warehouse. Historical data often has issues and understanding which weeks or months will impact the demand forecast is not always clear without a bigger picture perspective. For example, when one branch has a stockout, in some industries and for some customers, that will lead to a lost sale. In other cases, end customers may be happy to drive to a nearby branch that has the product in stock. Understanding your historical data and what it means for how closely sales reflect demand is a key step in developing an accurate demand forecast using historical sales data.
Our next step for our consulting clients after we have refined the historical data is to run demand forecasting algorithms. We run multiple, proven forecast algorithms within a math optimization model to determine if each SKU is trending up or down, and/or if there is a seasonality factor. These algorithms are optimized to maximize forecast accuracy. We can then add external factors within a machine learning model to account for additional macro or micro issues that impact the demand forecast.
The approach we take at Supply Velocity allows us to get the maximum benefit from proven supply chain analytics, data science and machine learning techniques. However, unlike data science focused consulting companies, we understand that the value from a demand forecast for a supply chain company comes primarily from how you use it.
Most of our clients choose a combined approach, using demand forecasting with inventory optimization. This approach allows them to go directly from a forecast to actionable recommendations such as reorder points and reorder quantities (or min/max settings). These recommendations are built on the SKU-level demand forecast, but also rely on techniques like ABC Inventory Classification, Safety Stock, and Service Levels to feed into a profit maximization inventory optimization model. For more information on our Inventory Optimization approach, please visit our service page here. Contact us to learn how we can customize our demand forecast and inventory optimization models to your business.