I’ve been writing about the Demand Signal Repository (DSR) market a fair bit recently (see, for example, “How to Best Describe Demand Signal Repositories”). DSR solutions leverage a variety of different types of downstream data, including POS data, to power more robust and dynamic forms of replenishment.
As the saying goes, “everyone talks about the weather, but nobody does anything about it.” Well, when it comes to supply chain management, most companies know about the bullwhip effect, but very few do anything about it. Dynamic replenishment, based on downstream data, finally offers a way for companies to “do something about it.”
When it comes using short-term forecasts for responsive replenishment, software solutions approach this in one of three ways. First, it can be done with flowcasting methodologies. In flowcasting, a store-level forecast by SKU is done. This is converted into a forecast that is propagated upstream from the stores and aggregated into replenishment requirements. This is how RedPrairie (an ARC client) has approached the problem. RedPrairie claims that a technology breakthrough allows its solution to efficiently handle the mountain of detailed data involved in this process. André Martin, the CEO of RedPrairie Collaborative Flowcasting Group, has worked with a major food company using this methodology. For related commentary, see “Eliminating the Bullwhip Effect across the Extended Retail Supply Chain.”
A second approach taken by some software vendors is to mimic store-level ordering policies. These solutions contain store order policy logic that anticipates what a store will order from the consumer goods manufacturer. One Network (a Logistics Viewpoints sponsor) is the primary practitioner of this style of responsive replenishment. One Network provides both the core data repository and supply chain applications in a software-as-a-service (SaaS) model. The data repository contains retailer and supplier master data, transactional data, and the stores’ replenishment policy information. Del Monte is a One Network client that has publicly discussed its use of this framework. The company is leveraging both Walmart and Kroger downstream data to make better replenishment and transportation decisions.
Terra Technology, who has had the most success selling short-term forecasting solutions based on downstream data, takes a third approach. Terra takes downstream data and uses pattern recognition. So, for example, for a large consumer goods manufacturer trading with Walmart, the Terra applications would have access to store inventory data, the store forecast of sales, distribution center (DC) withdrawals, DC inventory, and a DC’s open orders. The Terra solution looks at how to combine these data streams to produce the “most likely” forecast. One week out it may be that orders are most predictive. Two weeks out it might be DC withdrawals that have the most influence on the forecast. Three weeks out, it might be POS sales that have the most predictive power. These weightings can differ by product, and product weightings might also differ based on the warehouse territory they are located in.
Terra has been very successful with this approach. Procter & Gamble (P&G), after piloting at least one other solution (in which it played the role of coproduction partner) became a client in 2008 and publically stated that the Terra solution allowed it to lower its inventory by $100 million without adversely affecting service. P&G’s savings on this would be the carrying costs, probably between 15 to 20 percent of the $100 million every year.
In many ways, the DSR market is still immature. However, when it comes to using downstream data for dynamic replenishment, good solutions already exist.