The following is an excerpt from Enterprise Master Data Management: An SOA Approach to Managing Core Information, by Allen Dreibelbis, Eberhard Hechler, Ivan Milman, Martin Oberhofer, Paul van Run and Dan Wolfson. It is reprinted here with permission from IBM Press; Copyright 2008. Read the chapter below to learn why the authors say organizations need master data management (MDM) systems, or download a free .pdf of "Why do we need master data management systems?" to read later.
Chapter 1 is split into four parts. This is Part 2.
Table of contents
- Part 1: Introduction to enterprise master data management
- Part 2: Why do we need master data management systems?
- Part 3: What is a master data management system?
- Part 4: Business benefits of managed master data
- Chapter 2: Master data management as a service-oriented architecture enabler
- Podcast: Listen to a podcast with Dan Wolfson, co-author of this book
1.2 Why do organizations need a master data management system?
Master data has been around a long time, so why do we suddenly need MDM Systems to manage this kind of information? What makes this master data special? Why is it important? These are important questions—so within this section we look at how the information that is some of the most valuable information to an enterprise has become virtually unmanaged and often ungoverned. The fundamental purpose of an MDM System is to serve as the authoritative source for master data: An MDM System is a system that provides clean, consistent master data to the enterprise. If the business benefits of a managed master data environment are clear, then why is it that enterprises have unmanaged master data? Why do organizations have multiple, often inconsistent, repositories of data that should be maintained in common across the enterprise? For convenience, we call this data Unmanaged Master Data.
First, let's consider the distribution of unmanaged master data throughout a typical enterprise. Why are there multiple copies of master data? Are these copies redundant? This distribution may be viewed along any number of dimensions, including by line of business (LOB), by organizational change—such as mergers and acquisitions, and by the introduction of packaged software.
1.2.1 A Cross-LOB Perspective
Lines of business (LOB) are the natural segmentations of responsibilities that form within an organization, especially where the organization carries a broad portfolio of products or services. By their nature, lines of business often have unique perspectives on core business information, such as products (the products that this LOB offers), customer (the type of customer information that is important), and account (the nature of an ongoing relationship with a customer based on one or more products). For example, a line of business within a financial institution that is focused on deposits will usually carry different product information than a line of business focusing on investments—see Figure 1.1. Similarly, the customer information that is important to a mortgage department is often different from the information that is important in the support of checking accounts. In reality, neighboring lines of business experience varying levels of cohesion—some lines of business share a great deal of business information, while others share a good deal less.
Figure 1.1 Lines of business often maintain their own business information.
Within large organizations, lines of business frequently act as very different suborganizations, each funding and maintaining its own suites of applications and data. Even where similar, or indeed identical, applications such as Customer Relationship Management (CRM) or Campaign Management are in use, lines of business often install and manage their own instances of these solutions independently. For example, when multiple LOBs implement a software package like SAP, they often make unique customizations to the data models and look-up tables, which results in multiple independent implementations.
Typically, lines of business capture and maintain unique representations of core business information (customer, product, arrangement), each with their own unique slant on the usage and representation of that information. While regulatory requirements have a role in these differences, in many cases this issue is about control. A line of business sees this data as critical to its day-to-day operations and may not see value in sharing this data within the broader enterprise. All of these factors encourage lines of business to seek to control their own master data, and they sometimes act as barriers to the sharing of this business information.
1.2.2 A Cross-Channel Perspective
Each line of business may also have a number of distinct (distribution) channels to market. While these channels are often very similar, the resulting treatment of business information is often very different. For example, within a single line of business there are frequently entirely different solutions in place for attended channels (such as a branch office) and unattended channels (such as the Internet). These differences in customer interaction patterns across channels often drive a perception that the problem space differs sufficiently to merit an entirely different solution. In other cases, increased complexity is caused by evolution, with emerging channels adopting solutions that were simply not available when support for existing channels was defined.
Figure 1.2 Channel variance further scatters business information.
There are, of course, valid differences in business information across channels—Internetonly product offers, in-branch deals, and so on. However, these variances are often best viewed as just that—minor differences of the business information rather than fundamentally different information.
The location of master data can also be influenced by the realization that core information2 can be shared across lines of business, particularly where the adoption of a new channel acts as a catalyst driving a common view of master data across that channel but not across related channels. The result can be a unification of master data across lines of business for some channels but not for others. For example, many enterprises strive to provide a single point of entry for customer self-service over the Internet. Even if a customer has five different kinds of accounts at a bank (managed by five different systems), the organization will still want to present a unified view of that customer relationship through this channel. Indeed, one of the key business themes that we see across many industries is the desire to change the focus of the business from an account-centric one to a customer-centric one across the enterprise.
All of these factors add an additional dimension to the distribution of master data—the location of master data stores across lines of business, and distribution across channels, as shown in Figure 1.2.
1.2.3 A Cross-Business Subdomain Perspective
Different concerns across lines of business and channels often result in variation, not just in where and how business information is captured, but also in what information is required. In the case of customer data, the branch channel may seek to consider a much broader range of information (customer, contact preferences, contact and case history) than related channels such as partner networks, which may only be interested in a subset of this information. The result can be a fundamentally different scope of business information across channels or lines of business. Each channel and/or line of business may also distribute this information across solutions in different ways, adding further to the complexity of the distribution of customer information, as shown in Figure 1.3.
Figure 1.3 Variance in business information further fragments master data.
Additionally, different organization units may have different scopes of interest for product information. Financial controllers, for example, may wish to consider the profitability of different products, product groups, and services. This information is a different subset of product information than the information considered during the process of offering a product to a potential customer.
Variance of the business domain and the information needs of that domain drives further distribution of master data, because each line of business and each channel seeks to maintain its own unique perspective about the business information that best meets its needs.
1.2.4 A Cross-Application/Technology Perspective
An increasingly common reason for master data redundancy is the introduction of packaged applications and solutions (CRM, ERP, etc.). Typically, packaged systems are designed to manage their own master data. When multiple packaged applications are deployed in an environment, an interesting conundrum arises—each of the packaged applications will likely only store the information it needs for its own operations—so when you have two or more of them deployed in an environment, there is no common definition of the master data elements. For example, information about customers is normally needed by both an ERP system and a CRM system—because they each likely maintain unique customer attributes, neither represents a complete view of the customer information. As we describe in Chapters 3 and 5, a common pattern is to use the MDM System to support the complete representation of customer information through the aggregation or federation of customer data from multiple systems.
Integrating a packaged application system into an enterprise can be a difficult and costly endeavor, because the new system must be synchronized with existing sources of data, including master data. When packaged solutions contain multiple independent applications, they may need to synchronize master data within their own solutions as well as with the customer's environment. In an environment with many such applications, pair-wise synchronization can be complex and fragile. Using an MDM System as a common hub from which other systems are synchronized simplifies the number of connections and can improve the overall quality of master data and the manageability of the environment.
Finally, consider the effect of variance in the technical platform or application solution on the distribution of master data. Between lines of business or channels, many different representations of business information may evolve based on different platforms or applications. For example, if a customer has a well-tuned mainframe application already managing product data, extending that system to supply information to a new channel application may be perceived as too costly. Real or not, the perception is often that these platform differences are difficult to resolve, and often no attempt is made to integrate across different systems, which results in further distribution of unmanaged master data.
The result of all of these varying concerns is that master data is often widely scattered across the enterprise, with each channel, line of business, and solution stack evolving its own unique silo of master data. Where attempts to share business information do exist, they are usually ad hoc in nature and limited to a particular channel or product type. For example, it is not uncommon to find at least a couple of dozen stores of customer data in a financial institution.
1.2.5 Mergers and Acquisitions
Mergers and acquisitions serve to dramatically accelerate the replication of business information within an enterprise. Each party to a merger has its own distinct set of master data sources along the dimensions highlighted earlier—a sort of master data fingerprint. Without extensive effort to converge these data stores, the resulting merged organization will not be able to effectively leverage the combined assets (customers, products, etc.) of the new organization or be able to achieve economies of scale in the operation of the merged enterprise. For example, consider a case where organization A is to merge with organization B. Both organizations maintain LOB-specific stores of master data, as shown in Figure 1.4; however, both organizations consider themselves to have developed strong offerings for the Internet channel. Organization A shares information across lines of business within the Internet channel, while organization B has all but eliminated channel-specific perspectives on master data, and each line of business operates on the same data, regardless of the channel concerned.
Figure 1.4 Two organizations with different patterns of data distribution.
Merging these organizations yields a very different picture, however. The result is a dramatic increase in line of business-specific solutions, because each organization brings to the table its own solution, in each line of business. Within a specific channel (e.g., Internet), the two contrasting approaches of sharing information across lines of business, and eliminating channel-specific variances do not align well, which results in further complexity, as shown in Figure 1.5, with one solution seeking to be a channel-specific source of truth for all lines of business, and another seeking to eliminate channel-specific management of business information.
Consolidation and modernization of existing systems often require a similar kind of convergence. For example, an organization may, after several years of geographic growth, realize that each region has independently created localized systems that contain partially replicated and overlapping sets of data, which has led to an incomplete and inconsistent view of its customers, suppliers, and products. Addressing these business problems can be viewed as a merger of the different geographically based organizations and systems.
Figure 1.5 The resulting merger often suffers from the worst of all inputs.
In summary, there are many natural forces that have led many enterprises to have multiple copies of master data spread out across different lines of business, different communications channels, and different kinds of applications. As the number of unique copies of master data increases, synchronization via point-to-point connections becomes more complex, and the overall environment becomes more difficult to both manage and change. Indeed, when multiple systems manage the metadata, it is hard to achieve consistency of the information. For example, it is likely that many of the applications used to manage master data will have different rules for validating and standardizing the data—thus, even simple things like shipping address information for a customer may not be consistent.
More information about enterprise master data management
- Download the full chapter from Enterprise Master Data Management: An SOA Approach to Managing Core Information
- Read more master data management books in the chapter download library