Kim Dossey, Teradata’s manager for independent consultants, recently sat down with Jill Dyché and Evan Levy, founding partners of Baseline Consulting, to discuss master data management (MDM), enterprise data warehousing and their new book, Customer Data Integration: Reaching a Single Version of the Truth (Wiley, 2006), now available through major book retailers.

Jill Dyché and Evan Levy of Baseline Consulting have published a new book on CDI and MDM.

Kim Dossey: There’s certainly a lot of talk about master data management and customer data integration [CDI]. How would you define these two terms?

Evan Levy: In our book, we define master data management as “the set of disciplines and methods to ensure the currency, meaning, quality and deployment of a company’s reference data within and across subject areas.” This means applying standards and rigor to all of your company’s descriptive data. Customer data integration is simply MDM-specific to the customer subject area.

KD: Your definition implies that there are processes involved. But there are many vendors who label their tools MDM.

Jill Dyché: We see MDM as both a noun and a verb. In other words, it’s definitely a set of practices around managing enterprise data, encompassing critical activities like data management, data privacy controls and data quality processes. But it’s also a verb, in that MDM ultimately involves the automation of data integration and reconciliation.

EL: From a “process” standpoint, MDM encompasses work we’ve been doing in data warehousing for a few years. But from an automation and processing standpoint, it really is brand new.

KD: So you see a difference between MDM and data warehousing?

EL: Definitely. The data warehouse is a platform that serves as a repository of business-structured data—data that covers multiple subject areas that are integrated. It’s built to reflect the way that the company operates, and it usually stores historical detail. An MDM server—what we call a “hub”—is an operational system with a single focus: to cleanse, reconcile and integrate subject-area content from multiple systems—operational and analytical. MDM isn’t meant for query support; it’s focused on operational integration and the provisioning of accurate data to other systems. MDM is about infrastructure.

KD: What you’re saying is interesting, because many are claiming that MDM isn’t new.

EL: Well, it’s been around in the enterprise resource planning (ERP) world. The new aspect of CDI and MDM is that we’ve never automated the matching, merging, standardization, integration and validation of data. That kind of work has always been done either through ETL [extract, transform and load] or through specialized homegrown programming. But now, all that work can be automated.

KD: How then do CDI and MDM affect an analytic environment?

JD: CDI and MDM centralize data integration and reconciliation so that they occur once, not platform by platform. This simplifies transformation and integration processing within ETL and makes the data much better for BI [business intelligence]. This frees up resources to allow the addition of even more data to the data warehouse.

KD: Is that why the subtitle of your book is Reaching a Single Version of the Truth?

JD: Right. We see data warehouses as the single version of the truth for historical and multi subject area content. Customer data integration is the version of truth for operational access to customer data. It allows all the operational systems to reflect the current master detail at a particular moment in time.

KD: You’ve both worked in data warehousing and business intelligence for more than 20 years. Yet your book focuses on CDI.

EL: During that time we’ve seen the interest in customer data expand from BI to CRM [customer relationship management] to operational systems. In most IT organizations, the individuals with knowledge of customer data are in the data warehouse team.

JD: And it’s also because CDI is an easier pitch to management. The business case for CDI directly touches customers, and there can be huge ROI [return on investment] for customer-focused programs. Most of our clients are planning around MDM but starting with CDI.

KD: What are your thoughts on product information management [PIM], which is another critically important area for consistent master data?

JD: PIM is arguably more complex than CDI. A customer can have dozens of attributes, whereas a product can have hundreds. It’s also important to realize that product descriptions can vary greatly from company to company. We see CDI as a better on-ramp for a longer-term MDM strategy.

KD: Can you comment on Teradata’s recent MDM solution announcement?

EL: It’s interesting. We tell our clients with data warehouses to review and identify integration needs for their operational systems and their data warehouses. It’s important that MDM be enterprise-focused and not data warehouse-centric. Through some savvy partnerships, Teradata’s MDM offering packages its powerful database product with field-proven MDM processing functions. It’s really a great marriage that lets Teradata customers leverage their existing investments and resources.

KD: Jill, you’ve been presenting and writing a lot about data governance and data stewardship. These are really hot topics right now. What’s the link between MDM, CDI and data governance?

JD: Well, if you agree that data governance is really a framework for organizational oversight and policymaking around enterprise data, then it’s critical for MDM. Establishing definitions and rules around master data are key tasks.

KD: But they are for data warehouses as well.

JD: True, but for me, data governance is generally a luxury for data warehouses. For instance, a lot of Teradata customers have been really successful without having data governance committees or trustee councils overseeing data rules and outcomes. Data governance is a mandate for master data, where companies are working toward the goal of having common master data available to both transactional and analytical platforms.

EL: Teradata customers were there first. Because they were integrating large volumes of heterogeneous data before most companies, they got the policymaking and decision-making mechanisms in place earlier than other companies did.

KD: Looking ahead, how do you see CDI and MDM working together with the enterprise data warehouse [EDW]?

JD: Actually, if you take a CDI hub and let it feed an ETL process, it can exponentially improve the quality of the data on the data warehouse …

EL: … and save a ton of developer time. MDM is a boon to productivity.

JD: And people in the Teradata community have understood for a long time how good, clean, integrated data can bring about business improvements. T

Jill Dyché and Evan Levy can be reached through Baseline’s Web site at www.baselineconsulting.com.

Teradata Magazine-December 2006