Business Intelligence Best Practices - BI-BestPractices.com

Collaboration. Communication. Community.

 
 
 Printer-friendly
 E-mail to friend
  Comments
ADVERTISEMENT
Achieving Data Quality Through Master Data Management

by John Bair
As companies look to better manage their data, achieving a higher level of information quality is a key objective. Master data management (MDM) is an essential component that helps companies manage reference data across business functions and processes.

It’s worth revisiting data quality for a moment to emphasize that a comprehensive data quality program addresses data quality over a number of domains of responsibility. These domains include:

  • Executive sponsorship
  • Data governance
  • Data stewardship
  • Metadata and data lineage
  • Source control and release management
  • Data quality assessment and analysis
  • Data quality architecture
  • Reference data and hierarchy management
  • Data quality education

In a previous FlashPoint article called “Data Profiling and Quality Monitoring” (February 25, 2005), I noted that the earlier data quality issues are identified, the less costly are the remedies. This is true not only in the software development lifecycle, but it also applies in a broad sense to the enterprise information lifecycle. Data management professionals attempting to integrate transactions across systems often encounter inconsistent data from upstream systems that complicates or may even prevent cross-system rationalization of the data. This is where master data management solutions can help.

Master Data Management

In an article published in the August 2005 DM Review,“Master Data: The Linkage Between Business Functions and Business Processes,” Jonathan Wu defines master data as “the controlling set of values for reference data,” and explains that reference data contains the values that define the business context of transactions.

Although reference data must often be customized, organizations embarking on master data management programs are well advised to research applicable industry reference data and classification schemes sponsored by government agencies, such as ICD-10 (International Classification of Diseases, Centers for Disease Control and Prevention), and non-governmental consortiums, such as ISO 3166 (Country Names and Code Elements, International Organization for Standardization). Research into industry reference data initiatives may also help bring data quality professionals into contact with business sponsors within the organization.

In addition to code sets or industry classification schemes, master data management is equally applicable to the key data entities that provide context for the transactions of a corporation, including but not limited to:

  • Customer/Account/Subscriber
  • Product/Item/Service
  • Vendor/Account/Supplier
  • Legal Entity/Organization

Effective MDM solutions can improve data quality early in the information value chain, and can deliver direct business value to all participating systems, whether the systems are transactional or analytical. Examples of the benefits of master data management can include:

Business Application Potential Benefit of Master Data Management
Risk Management Reducing exposure to business partners that trade through multiple purchasing systems or business units
Purchasing Reducing costs by aggregating demand or purchases
Inventory Management Reducing inventory by identifying similar products from different manufacturers
Reporting Simplifying data integration and standardizing dimensions to improve report consistency

Table 1

The exercise of identifying the potential uses and benefits of MDM can help data quality professionals find business sponsors to quantify the benefits and to help champion the program.

One Size Does Not Fit All

The market for MDM products is growing as vendors respond to customer demand for better master data management tools. Reference data products are offered by many data providers; software product offerings include specialized MDM solutions as well as ETL, EII, and database management solutions that are being re-branded or reinvented into master data management products.

As you work to identify and manage master data, you will discover that master data management solutions tend to be complicated by the way that master data spans multiple applications and business units. Some common issues are described in the following table:

Business Issue Implications or Complications
Standardization Local variation is often required to accommodate legitimate differences by business unit, geography, or function.
Centralization Ownership of various attributes or aspects of master data entities may need to be divided between different applications or business units.
Change Management Cross-application versioning or reference data management may be required to manage master data that was previously managed independently. In addition, ownership may change over time due to business or system transformation.

Table 2

Each organization has different tolerances and constraints surrounding these issues. The MDM solution must be flexible to accommodate change and varying degrees of system and business readiness and participation.

Master Data and Service-Oriented Architecture

Many organizations are implementing service-oriented architectures (SOA) in an effort to integrate and augment legacy applications with new functionality or as part of enterprise application strategies. A primary advantage of SOA is that the architecture provides a means to loosely couple multiple interacting but heterogeneous applications. Service contracts or application programming interfaces (APIs) help to shield individual consumers and services from change; therefore, SOA can provide opportunities to migrate away from legacy-independent reference data toward MDM services. The following figure shows a sample SOA implementation of an item master data service:

ItemMasterService

It should be emphasized that no service-oriented architecture or messaging standard can auto-magically resolve the semantic disconnect between heterogeneous applications. Therefore the SOA-MDM relationship can be thought of as symbiotic, since MDM services can help bridge the semantic gap between applications in a service-oriented architecture.

Master Data and Data Warehousing

Reference or master data is frequently translated into dimensions in data warehouses, marts, and analytic applications. This data is rarely simple, and is often organized into hierarchies and other classification schemes.

The dynamic nature of the business and information systems causes master data to change over time. Since master data provides the context for transaction data, and since the data warehouse manages history, it is usually insufficient to maintain only the current state of master data, because doing so may lose the original context of historical transactions.

For example, suppose Amy, a West Coast sales representative, earns credit for a customer sale in July. Later, in August, the sales department reorganizes and Amy is transferred to the central sales region. If the historical sales department reporting relationship is not maintained, subsequent sales reports for July may not correctly count sales by region.

The information architect may choose for the master data service to maintain history, or she may choose to capture changes and manage the historical master data separately in a data warehouse. Either way, it is important to explicitly decide how historical master data should be managed.

Master Data and Governance

As noted earlier, master data management solutions are complicated by the way that master data spans application and business unit boundaries. Because of this, master data management programs are often implemented hand in hand with business data governance initiatives.

Lack of an effective data governance program is often cited as an excuse for lackluster master data management in an organization. However, it is not enough for data management practitioners to throw their hands up in the air or to passively wait for the business to start or revive a data governance initiative. Likewise, it may not be realistic to expect that all of the various IT and business units will ever agree to completely trade in autonomy for centralized master data management. Often there are legitimate regulatory and business reasons for maintaining local variation. What is needed, then, is a sensible plan for enterprise standardization (where possible), local variation (where necessary), and a flexible system for managing enterprise data and mapping sensible variation.

Achieving Results

Given the cross-system and business-unit nature of master data, a comprehensive master data management program brings together the business leaders, application architects, and information architects to create winning MDM strategies and solutions. Although this requires an investment, the resulting successful master data management program can deliver tremendous direct and wide-ranging business and IT benefits to the organization.

John Bair -

John is Chief Technology Officer at Ajilitee, a consulting and services firm that specializes in business intelligence, information management, agile analytics, and cloud enablement. Johnís technology career includes leadership positions at companies such as HP, Knightsbridge Solutions, and Amazon.com. He has decades of experience building complex information management systems and is an inventor on six data management patents.††

Editor's Note:†Find more articles and resources in John's BeyeNETWORK Expert Channel.†Be sure to visit today!