Supply chain analytics and manufacturing business intelligence (BI) require substantial work just to collect and organize the data because so much of it comes from external sources such as subcontractors, suppliers, and syndicated data services. This variety of data sources can make supply chain data management a challenge.
The software then used to analyze the data (see sidebar) will depend mostly on the complexity of the problems a manufacturer is trying to solve, and whether it has the technical resources and analytics know-how to do the job in-house, according to several experts.
“The major foundational thing is an informational model that represents your supply chain and trading partners,” said Bob Parker, group vice president at IDC Manufacturing Insights, a research firm based in Framingham, Mass. “That’s typically ERP [enterprise resource planning].”
But not all ERP packages make good foundations for supply chain analytics. It depends on which of two informational models the ERP predominantly uses, Parker said. The transactional model derives from ERP’s historically core functions and centers on double-entry bookkeeping. In contrast, analytical data models were adopted by vendors of supply chain management (SCM) software to address the integration, planning, and execution issues inherent in supply chains.
ERP’s predecessor, manufacturing resource planning (MRP II), had analytical features for planning and optimizing resources, but it wasn’t very good at handling multiple contract manufacturers or fluctuating demand, Parker said. ERP vendors responded with advanced planning and scheduling modules that used analytical data models to help manage more complicated supply chains with multiple constraints. They later added BI, analytics tools, and online analytical processing (OLAP) database technology. More recently, in-memory analytics is speeding up performance.
“It’s taken [ERP vendors] a while to perfect it, and they’ve done a good job,” Parker said. “What we’re seeing is the ERP guys moving up and competing with best of breed.” Even so, ERP vendors lag behind the SCM specialists in addressing vertical industries, nor are they as good at handling “foreign” data coming from outside a manufacturer’s ERP, Parker said.
Why SaaS supply chain analytics makes sense
Nowadays it seems Software as a Service ( SaaS) is taking over every category of software, but the cloud-based deployment model may be especially well suited to solving the data integration and training issues of supply chain analytics.
“Software as a Service allows us to move quicker,” said Lora Cecere, a partner at Altimeter Group, a San Mateo, Calif.-based analyst firm. “The average businessperson doesn’t really have the acumen to build and develop the models that need to happen for predictive analytics.” Furthermore, people with the exceptional statistical skills to develop and use predictive analytics have limited career paths in most companies, but can thrive at SaaS analytics providers. Better for them to build the models so business users can access them online, Cecere said.
As a compute- and storage-intensive, batch-style process, analytics can benefit from SaaS economies of scale, according to Parker. “Why have all that hardware on site just to use it for eight hours? With SaaS, I only have to pay for the CPU cycles when I use them.”
SaaS also encourages the data cleansing needed to build enterprise data models by forcing companies to conform to the provider’s data format, Cecere said.
But that doesn’t have to mean one-size-fits-all conformity. Parker said SaaS analytics specialists have started to offer “wrappers” that allow manufacturers to customize the software’s rules engine to handle the highly specialized calculations that used to require homegrown analytics.
SaaS analytics providers usually spare manufacturers the hassle of integrating suppliers and customers into the supply chain planning network, allowing for a more collaborative planning process, according to Parker. The retailer Best Buy, for example, could use analytics-based planning to coordinate production and distribution with notebook manufacturer Asus and one key component supplier, Intel Corp. “I think you’re going to see more of that, but we don’t see a lot of it yet,” Parker said.
SaaS server farms could also enable what Cecere calls “sense and respond” analytics, which she said could be the next generation of analytics. “Models are complex and require horsepower,” Cecere said, but SaaS enables calculations to be spread across servers and processed in parallel.
Analytic “systems of engagement”
Basic BI actions such as reporting are reactive in nature, though ad hoc querying and discovery are somewhat proactive, Cecere said. Predictive analytics, in contrast, is what will enable manufacturers to use cloud-based BI platforms to formulate intelligent responses to fast-changing situations, becoming sense-and-respond organizations.
Supply chain planning already has some predictive analytics, but most of the optimization it provides is deterministic: It assumes events are caused by preceding events and fixed laws, the way linear programming did in the early days of computing. “It doesn’t really deal with variability,” Cecere said.
To be able to predict and respond to variability, manufacturers will have to build something that doesn’t exist now: an enterprise data model. “Without an enterprise data model, we can’t connect these models together,” Cecere said.
Manufacturers will also need to extend to their sales and design departments the same tactical and strategic planning technology many of them have been using for decades on materials planning and production. By applying more analytics to sales and operations planning (S&OP), revenue management, and product portfolio management, they will make better decisions about which products to sell, Cecere said.
Software vendors are only just now discussing the concepts, and few have any real plans on the drawing board. “We have not been able to develop a more holistic planning tool,” Cecere said. Bottom line: it could take five years for sense-and-respond analytics to become a reality.
Regardless, manufacturers might not even be ready for it. Cecere said most that own advanced planning and scheduling software don’t use it, primarily because their employees never developed the skills.
Parker envisions a similar evolution from reactive analytics to a more collaborative and responsive approach. “The last 10 years we’ve been investing a lot in BI and analytics tools trying to create a system of decision,” he said, and manufacturers have focused on creating scorecards to measure supply chain performance against industry benchmarks.
The next phase, Parker predicts, is a “system of engagement” where manufacturers enlist their global partners in applying predictive analytics and root-cause analysis to manage supply chains collaboratively.