How to Manage Product Data in a Multi-Enterprise World
Product content and data management (PCDM) involves a set of related disciplines, technologies, and solutions used to create and maintain consistently interpreted product data and to facilitate commercial exchange. PCDM will become increasingly important during the next five to 15 years as enterprises seek to eliminate waste and create competitive advantage. Although the business drivers for PCDM are clear, the technical underpinnings are not as well defined. We consider product registries and business-to-business data synchronization to be the likely technical responses to the PCDM challenge.
The PCDM Challenge: The Registry
Because PCDM is a cross-application, cross-departmental, crossfunctional, and cross-enterprise activity, whole new classes of applications will begin to gain traction during the next five to 10 years. Furthermore, established applications will require an extensive overhaul (including a redesign based on new architectures) to contribute to PCDM, and these changes will be leveraged by new supply chain management solutions.
For example, typical enterprise applications assume that product data persists behind the firewall and that this data pool is based on item hierarchies or item numbers. Based on the heterogeneity of product data created by the exchange of information within and between trading communities, this is an insufficient approach. For those industries that seek to create shared external data pools (based on a distributed registry model), product data will be anything but static, and it will not be subject to any single enterprise's classification scheme.
In the case of a peer-to-peer-based registry model, buyer systems would be required to acquire updates and new product data from external systems. Furthermore, seller systems will have to ensure that external systems are fed and maintained by several internal systems, such as those related to product definition (engineering), sales and marketing, and even logistics systems for item availability. As a result, PCDM systems must support reconciliation with external systems, driving the creation of new applications or services. To maintain consistency in data across enterprises, the following PCDM services, facilities, and capabilities are likely to evolve across industries during the next decade (see Figure 1).
B2B Value-Added Applications and Services These capabilities will be required to sustain the underlying technical infrastructure that is central to registry-based PCDM. In the same way that transaction data has been exchanged for some time via value-added network services such as in electronic data interchange VANs the exchange of product data will be facilitated by service organizations. B2B valueadded services will provide the trading community with management and support for variants of industry standards within trading communities. These services are required to support the bidirectional exchange of product data with multiple internal applications, as well as external trading partners.
In addition, traditional enterprise applications such as enterprise resource planning II, supply chain execution, customer relationship management, supplier relationship management, partner life cycle management, and new applications focused on multi-enterprise business processes will evolve to accept product data and change notifications from external enterprises and not just manage the internal processes for which they were used previously.
Messaging, Update, and Synchronization A scalable PCDM messaging infrastructure will require bidirectional notification of inaccurate or expired data. Mission-critical buy or sell processes will require real-time processing. As a result, such processes will become a critical requirement for the real-time enterprise. Notification will be generated from the primary data pools and sent to registered consumers of such data, directly (in the case of P2P networks) or via centralized data registries. These services are fulfilled by enterprise application integration vendors, although, as industries consolidate around multiple schemas or adopt singular-registry models, these services are likely to evolve to a specialized domain area, because the requirements will be high-volume, small-content, and real-time asynchronous messaging.
Data Pooling, Cleansing, and Normalization These three steps are grouped because the technologies are combined when enterprises build Web-based catalogs.
Data pooling is the gathering of all necessary product data into one central location for a single enterprise (if large enough) or for multiple enterprises in a third-party solution, such as an e-marketplace or hosting service. (The necessary data is likely to be a subset of the total data associated with a given product.) This step involves the repetitive extraction and loading of heterogeneous data types of multiple formats from multiple systems into a common repository. Several technologies will continue to improve the efficiency of the process, although it will never be free of human labor. Other technologies that affect the maturity and evolution of this stage include artificial intelligence search capabilities, natural language recognition, image recognition, 3-D bar coding, radio frequency identification, and the expansion of network, storage, and computing capacity. However, all enterprises that decide to synchronize their product data through methods other than via proprietary and traditional point-to-point EDI will have to create formal data pools.
Cleansing and normalization involve the laborious process of homogenizing diverse data into a consistent structure with uniform units of measure. Circa 2015, the value of this function will create whole swatches of semi-automated offerings to parse pooled data and prepare it for registration. Cleansing and normalization will result in the final preparatory stage by which data is completed and made compliant to specific standards; data gaps will be highlighted for filling in; and errors in data format and content will also be highlighted. Exception processing becomes the most-valued feature, although manual labor remains to resolve errors. This will grow to support diversity in the sources and pools of product data within the enterprise and support propagation to internal applications and trading partners.
By 2013, original equipment manufacturer extraction, transformation, and loading tools will be combined with native decision support and self-learning capabilities to dramatically reduce the manual labor required to load and validate item data.
Large enterprises may need to establish these data pool environments internally to compensate for the heterogeneity born of divisional autonomy and application heterogeneity. In fact, the entire ecosystem will be comprised of numerous certified data pools.
Compliance and Endorsement Compliance ensures that participant product data is in conformance with accepted industry standards. This is an independently verifiable stage. Compliance must also support individual enterprises' quality assurance processes for inbound and outbound product data; however, the methods employed to do this will diversify, given the variability in relevant data volumes and business criticality. For strategic categories, this will be among the most-highly valued, as domain-specific applications persist in retaining control over portions of the product data life cycle. Over time, compliance will begin to exhibit bidirectional capabilities, as the range of PCDM capabilities extends to diverse data types that enrich product understanding. Requirements will be based on application type (for example, historical analysis vs. inbound logistics), diverse data types (such as RFID), manufacturing process definition, and item maintenance histories.
Product Registries Like a telephone directory, a registry acts as the main source for users and automated applications, enabling them to find products, determine the owners of products, and locate where products are being bought and sold. At the highest industry level, the registry will be made up of multiple registries perhaps distributed by country all connected to a single, large virtual registry used to synchronize them all. On a small scale, a single enterprise may replicate this functionality within its PCDM application stack behind the firewall, so that its other departments and applications can ensure that all product data is kept synchronized around the enterprise.
Multiple registries will continue to develop as shared resources among trading groups, especially those based on common items within a particular industry. Although multiple industry or trade registries will be active, the distribution of product data and PCDM functions will be highly diverse. Thus, there will be diversity in the ways that registries are created, owned, and managed. Diversity will be a product of variability in the volumes of data necessary to sustain synchronization, the business practices, and cultures associated with given categories, variability in enterprise buy-sell strategies, and trading-partner power positions.
In a few cases, some high level of cooperation among industry participants will create the need for distributed, local registries connected to one global registry. This is the case with CPG/retail and the efforts under way with UCCnet and driven by global commerce initiative. In these projects, the physical, logical, performance, and economic limits of postmodern computing infrastructures will prohibit the ubiquitous adoption of a highly centralized directory structure.
The era of treating product data as a departmental or enterprise matter will be drawing to a close, making way for the cross-enterprise adoption of PCDM. Users have been building data warehouses and deploying ERP as part of the concept of a single source of product data, as opposed to multi-enterprise, synchronization-dependent modes of product data management. Although internal management of product data is a nagging issue, users should begin to focus on and experiment with addressing the problem from a multi-enterprise perspective.
Bottom Line
Product content and data management is beginning to be recognized as a vexing and important multi-enterprise business-to-business challenge. The technology needed to manage product data inside and among enterprises is immature, but it will evolve and emerge in complementary stages, partly application-based and partly services-based. Enterprises should monitor the adoption of the PCDM technologies and services and begin building the skills necessary for selective adoption as they mature.
