Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

May 13, 2003

Supply Chain Intelligence

Know what to expect and how best to achieve it

by Ram Reddy

Supply chain intelligence (SCI) of late has become the "new new thing" in the area of supply chain technologies. SCI technologies promise to extract and generate meaningful information for decision makers from the enormous amounts of data generated and captured by supply chain management (SCM) systems.

The focus of SCM technologies primarily has been on providing operational and transactional efficiencies in the areas of sourcing, manufacturing, and distribution activities within a firm and across its supply chain. (See "The Evolution of Supply Chain Technologies," Jan. 14, 2002.) Applying the concepts of business intelligence to data from SCM systems, SCI technologies seek to provide strategic information to decision makers. Information categories range from what-if scenarios for reconfiguring key functions in sourcing, manufacturing, and distribution to measuring the ability of a supply chain to produce cost-effective products. Even before the term SCI was coined, data collected across the supply chain was crunched, numbers were analyzed, and information was generated for decision makers to reconfigure supply chain functions. Technologies ranging from mainframe-based multidimensional spreadsheets to PC-based statistical analysis tools did the heavy lifting for pre-SCI analysis. The biggest challenge in building these supply chain analysis frameworks was in aggregating data from multiple sources.

The Big Challenge: Data Integration

The data integration challenge for SCI technologies has become increasingly complex with the growth of popular application packages for ERP, SCM, and product life cycle management. Data from these multiple sources needs to be extracted and transformed to a common format before it can be used for any SCI analysis. Compounding the complexity of this data integration challenge is the need to incorporate the huge quantity of structured and unstructured data that a modern enterprise produces and receives.

Structured data is typically generated from transaction-oriented systems. It could be transmitted via EDI, XML, or plain ASCII text files. The typical data feed contains standard information from factory floor, supplier, customer, or logistics systems. Unstructured data, conversely, will comprise items such as terms and conditions of a supplier or customer relationship and other contractual information that may not adhere to a standard format.

Third-party research data — market trend analysis or survey data — is typically incorporated into SCI systems for analytic purposes. Scrubbing the data from multiple sources into a common format is not a trivial challenge. A major research effort in tackling this data integration challenge is being spearheaded by the Advanced Database Research Group, Department of Management Information Systems, at the University of Arizona. The biggest challenge to realizing the benefit from integrated applications in areas such as CRM, ERP, and SCM has been the difficulty in integrating data from other sources. The data integration challenge for SCI is even more complex because it vacuums data from all data sources within an enterprise and across the supply chain. (See "The Cost of Integrity," March 8, 2002.)

Good Enough Data Integration

While data integration is and will remain a major challenge in implementing SCI technologies, trying to solve that problem will sidetrack an organization from realizing the benefits of an SCI application. In my experience, tackling the technology problem of data integration has sidetracked not only many BI efforts but also SCI efforts. Solving the data integration problem of data aggregation across the supply chain is like trying to nail a target moving at hyperspeed. The data to be integrated is a representation of data extracted from the source systems at a point in time. The source systems themselves are changing and generating new data constantly. Trying to extract, transform, and synchronize the data from these multiple sources is a tough problem to solve and can suck up IT development resources.