Reduction of Safety Stock without loss of service
SKUs which exhibit “noisy” demand behavior (display significant deviations) tend to receive relatively high safety stock levels. These high levels of safety stock are partially caused by the inability of traditional statistical tools to distinguish between abrupt changes and instability in systematic patterns in the data such as a repeating sale/inventory life cycle. The resulting extra “cushion” of inventory can amount to millions of dollars in losses that would have been prevented with a more accurate prediction tool.
CB4’s solution analyzes the demand patterns of SKUs at a store/POS/warehouse level in order to determine when the demand for an SKU has changed. In many cases, even within SKUs that exhibit “statistical noise”, CB4 can identify consistent recurring patterns which allow a more accurate recommendation of the right safety stock level per SKU at each location. CB4’s solution is data agnostic and can receive additional data that might impact the demand patterns of the SKU. These recommendations accumulate to a significant reduction of safety stock levels, which could amount to millions of dollars. At the same time, the use of CB4 pattern-identification can prevent out-of-stock situations and assure a consistent service level to the customers.
Forecasting adjustment by pattern Demand Sensing
Inaccuracy in forecasting causes loss in revenues along the whole supply chain and complicates operations significantly. A key factor in the inaccuracy of traditional forecasting tools stems from the fact that they are simply not granular. Conventional statistical forecasting tools work top-down: they generate models based on accumulation of units at a higher level and then break down the accumulated forecasts to an SKU/store level by relying on their proportion. The result is a forecast that might be very accurate at the supply chain high level, yet very inaccurate at a granular SKU/store level, where it is practically needed. For example, forecasting accuracy of an SKU might be over 95% at the main warehouse, while deviations of ±25% are often found at stores and local warehouses.
CB4’s solution models and analyzes the patterns of demand for an SKU at a store/warehouse level in order to determine when the demand changes. The behavior analysis of each SKU at each store is done independently. When a change is detected, the conventional time series forecasting can be adjusted automatically. CB4’s solution is data agnostic and can receive any data that might impact the demand patterns of a product.
As retailers expanded their activities to online commerce, over the years many developed two separate channels of sales: online (internet) and offline (in stores). Although it is clear that the customer experience and behavior is affected by both channels simultaneously, most of the existing analytics tools tend to be platform specific - either operating on ecommerce data or on retail in-store data. Such an artificial separation prevents the retailer from understanding the key drivers that can lead to positive or negative sales impacts between the channels.
Being data agnostic, CB4’s solution can receive data from stores (e.g. sales, promotions, locations, prices, etc.) and simultaneously obtain valuable insights from retailer’s online activities (e.g. page views, impressions, coupons, clicks etc.). All that is required from the retailer is to decide on which target variable to focus the analysis – sales, profit, market volume, etc. Once the target variable is selected CB4’s solution generates AUTOMATICALLY the key driving patterns that affect the target and related recommendations how to use them. The patterns can then be grouped by any hierarchy and sorted by their significance and by their expected revenue increase. Being granular, the analysis can be as specific or as general as a user desires. For example a user might be interested in understanding the effects of online sales on a specific product in the stores (or even in a specific store) or investigate an entire product category.