Installing the building blocks for manufacturing business intelligence (BI) can feel like a lot of heavy lifting. After all, components such as master data management (MDM) systems and enterprise data warehouses represent some of the most expensive and complex IT projects many organizations undertake.
All this up-front data management work is required because manufacturers are swimming in ever larger volumes of highly detailed data, and that’s not going to change. Research firm Gartner Inc. (Stamford, Conn.) estimates that enterprise data volumes will grow 650% in the next five years.
Data volumes aren’t the only issue. It’s also waves of information coming from a variety of sources. “We’re seeing the convergence of the physical, the digital and the information supply chains,” said Jane Barrett, vice president of supply chain research at Gartner. “Along with all the traditional sources, such as ERP systems, there is point of sale [POS], remote monitoring of equipment and now even social media.”
This large and constantly updated flow of data can create severe data consolidation and synchronization headaches. Fortunately, there’s a payoff for the pain when this foundation supports timely forecasts that are based on the latest information from supply chains and the marketplace. And that can be the key to attaining the elusive data-driven supply chain and realizing the promise of lower inventory costs, increased sales and a sharper competitive edge for capitalizing on new opportunities.
Understanding demand signal repositories
The key is fine-tuning data management systems so they not only collect relevant data, they consolidate and integrate it for analytics systems. The demand side of the supply chain planning equation has traditionally received the most attention. After all, it’s a highly volatile area that in recent years has seen a variety of options for collecting POS data and establishing demand-signal repositories (DSRs) to standardize and manage the relevant data.
DSRs are a variation of data warehouses for organizing large volumes of information, including store- and SKU-level records, and integrating everything into a coherent whole that accommodates analytical applications. “The demand signal repository is key both for understanding all the different types of data available and for helping you decide what to do with it,” Barrett said.
DSRs create a foundation for demand-sensing software that flags key business indicators, such as inventory levels, stockouts and spikes in buying patterns that presage an emerging sales trend. DSRs, along with good MDM practices and the right analytics, can result in significant reductions in inventory costs through better inventory turns. The combination also aids sales and marketing managers with the information they need to evaluate the success of promotions and other demand-shaping activities.
In the future, DSRs available as Software as a Service (SaaS) solutions could increasingly help jump-start the data management and consolidation efforts associated with supply chain number crunching, especially for smaller, resource-constrained manufacturers.
The evolving role of supply signals
While demand-side analytics grab a lot of attention, analysts warn that focusing entirely on demand may blind manufacturers to other important elements of supply chain analytics. A better strategy is to also look closely at supply signals.
”There has been a lot of interest in the demand-driven supply chain, but supply can be just as volatile in some markets,” said Simon Ellis, practice director for global supply chain strategies at IDC Manufacturing Insights, a research firm based in Framingham, Mass. Ellis points to uncertainties in the high-tech industry resulting from recent supply chain disruptions in Japan.
The good news is that integrating supply data doesn’t require fundamentally different technology than what’s used for demand data. “I argue it’s all about having a data repository, so it doesn’t matter if it’s demand, supply or operational data,” Ellis said.
Modern data warehouses can use logic for detecting data inaccuracies, including information that’s not in the right form or numbers that are out of the logical range that is expected for a specific category. Added to that are embedded analytics, which enable companies to slice and dice data without having to send it out of the repository to a separate BI system. Eliminating this step when dealing with extremely large data sets not only saves time, it reduces the drain on the networking infrastructure.
Manufacturers are using supply data in more sophisticated ways, such as attempting to anticipate the future. For example, the maturing discipline of predictive analytics can combine statistics, artificial intelligence and data management to help a firm navigate unexpected supply shortages, Gartner said in a new report, Next-Generation Supply Chain Predictive Analytics: A Cornerstone to Demand-Driven Value Networks. By analyzing each customer’s tolerance for late shipments, the company can then decide who to serve first and who will more easily deal with delays.
Gartner calls this a game changer. “It fundamentally transforms the supply chain management model that was based on aggregate information, averages and generalized models into a tailored response based on the unique characteristics of customers, products or suppliers,” the report said.