Supply chain analytics appears to be a poorly understood technology in dire need of some best practices. According to a study released early this year by
The research firm surveyed more than a hundred manufacturing and service companies and found the use of analytics in most of them to be ineffective. More than two-thirds expressed a need for simpler analytics and metrics, and half were dissatisfied with the process they use to create analytics. Only a third of the companies were happy with their analytics technology; the rest complained about ease of use and data quality.
Ventana also noted a large gap between the expectations of senior executives and lower-echelon employees, which suggests a need for better alignment of business goals for BI and analytics. For example, the plant profitability metric was undervalued by middle managers and employees compared with executives. Similar disparities appeared in customer satisfaction and inventory metrics.
Ventana advised companies to begin replacing spreadsheets—overwhelmingly the most popular tool, but one whose effectiveness companies usually exaggerate—with dedicated BI and analytics applications.
“Generally, you need to keep things as simple as possible, but not so simple that they become simplistic,” said Robert Kugel, Ventana’s senior vice president of research. Companies must also build internal skills and make sure the necessary data is available, he said.
In an earlier BI study, Ventana recommended seven other best practices (see sidebar).
Planning an effective supply chain analytics strategy
Supply chain analytics and manufacturing BI raise cultural and organizational issues that strongly affect how companies apply the technology, according to James Kobielus, senior analyst at Forrester Research Inc., based in Cambridge, Mass.
“Decide what kind of organization you actually are,” Kobielus advised. “Are you an organization that goes by the numbers, or are you more qualitative? If you’re more the latter, analytics is not necessarily something you go deep into.”
Manufacturers can vary widely in their need to make analytics available throughout the enterprise. Some may have departments like marketing or human resources running analytics; others might want to limit analytics use to the chief financial officer and highly trained statisticians in operations, for example.
“Figure out which are the most appropriate groups in the company to roll out the analytics tools to,” Kobielus said, then categorize them as either casual users or power users. “Figure out who your core users will be, and what specifically do they need. Some just need reports and dashboards.” Power users, on the other hand, need advanced tools such as what-if scenarios and predictive analytics. “Quite often, these are people who are involved in analyzing different scenarios, and they’re often at the corporate level,” where potential partnerships and acquisitions are evaluated, he said.
The next step is to decide how deeply into the data the different user groups need to go. “Once you’ve established all that, figure out how you’re going to empower their needs,” Kobielus said. Casual users can be satisfied with an easy-to-use, self-service BI tool that runs on ERP or another enterprise application, but analysts in the operations research department will still need their high-end analytics suites.
Once the user application has been selected, you still need the right “plumbing” underneath, Kobielus said. That means establishing an analytics database with a schema on top to match, harmonize and cleanse the data. There should also be data-governance policies to ensure the trustworthiness of back-end data sources.
The cost of all these infrastructure improvements can quickly get out of hand, which is why companies should carefully scope out lowest-cost solutions. “More and more of the BI that we see out there in small to midmarket companies is being provided by outsourcers,” Kobielus said. Such Software as a Service (SaaS) providers often provide the prebuilt integration to supply chain partners and data sources that can quickly eat up IT budgets when handled in-house. “You get a good, solid, basic BI environment that probably serves 90% of your needs,” he said.
While more companies seem to be outsourcing their BI and using self-service tools, that’s not the case for those with highly specialized needs, according to Kobielus. “A lot of the nouveau stuff like predictive analytics is being left in-house because it’s being used by a small cadre of users,” he said.
Kobielus said some SaaS offerings that specialize in vertical industries provide another means of inculcating best practices without having to build them up internally—an effort typically beyond the resources of small manufacturers.
“They know their core businesses, but they don’t know anything about building statistical models,” he said. “The outsourcer is essentially a center of excellence.” Inventory management and logistics are common specialties of SaaS analytics providers that are geared to manufacturers.
Kobielus said he and Forrester analyst Boris Evelson researched the benefits of forming a business intelligence solutions center (BISC) to cultivate BI best practices. They said a BISC goes beyond similar approaches such as BI competency centers and centers of excellence by being governed by the business side rather than IT, and by focusing on business solutions instead of data warehouses and other infrastructure components.
Implementing a companywide analytics strategy can prompt negative reactions from some quarters, according to Kobielus. He advised proponents to be ready to answer skeptics who will ask why existing tools aren’t adequate and to take steps to address resistance from analytics specialists, who are accustomed to working independently in departmental silos.
“There will be cultural pushback from those teams,” Kobielus said, and they might react negatively to sharing expertise with other analysts or outsourcing their skills to a service provider.