106 Part I • Information Technology
data changes are broadcast through a central service to all
subscribing databases. Redundant data are kept, but there are
mechanisms to ensure consistency, yet each application does
not have to collect and maintain all of the data it needs. In the
persistent approach, one consolidated record is maintained,
and all applications draw on that one actual “golden record”
for the common data. Thus, considerable work is necessary
to push all data captured in each application to the persistent
record so that it contains the most recent values and to go to
the persistent record when any system needs common data.
MDM supports all uses of data, from operational to
business intelligence. In good MDM, there is no delay for
any application in knowing any fact about master data. For
MDM to be successful, an organization must create a
strong data governance process, often including data stew-
ards. We will describe data governance later in this chapter.
MDM requires a discipline around managing data.
But, does MDM pay off. The general benefit is that the
whole enterprise works from a single version of the truth
about key organizational data. This consistency and only
planned redundancy reduce errors, misunderstandings, and
wasted efforts to reconcile differences across business units
and with stakeholders. Also, the impact of changes of key
data values or even data models can be huge when master
data are not carefully managed. MDM also greatly simplifies
satisfying the data quality requirements of various regulations
such as Sarbanes–Oxley, HIPAA, and Basel II (Russom,
2006). Here is an example of the logic to justify customer
master data management and a strong program for data gov-
ernance: Consistent data drive better data matching, better
matching drives better customer identification and modeling,
better identification and modeling drive better customer inter-
actions and campaigns, better interactions and campaigns
yield higher hit ratio, and higher hit ratios result in more rev-
enues (Dyché, 2006). The root of this chain is MDM.
The Data Management Process
A manager of real estate, personnel, or finances is familiar
with the basic but essential functions necessary to manage
effectively those resources. Figure 4.6 lists the generic func-
tions for managing any business resource. This section
examines each of these functions within the context of data
management. An important point to note is that, as with
other resources, every business manager should be involved,
in some way, in every one of these functions for data.
1.Plan Data resource planning develops a blueprint
for data and the relationships among data across
business units and functions. As with most plans,
there will be a macro-level data plan, typically called
an enterprise data model, to identify data entities and
relationships among the entities and more detailedplans to define schedules for the implementation of
databases for different parts of this blueprint. The
plan identifies which data are required, where they
are used in the business, how they will be used (i.e.,
what they will be used to produce), and how much
data are expected. This plan must then be communi-
cated to all business functions that are involved in
aspects of data resource management. For example,
system capacity planning must be informed of this
schedule, along with data and processing volumes, so
that adequate computer and network technology can
be in place to operate and access these databases.
2.Source Decisions must be made about the timeliest
and highest-quality source for each data element
required. For example, should customer sales data be
collected at point-of-sale or entered later? Concerns
over error rates, frequency of changes, chance of lost
paper documents, technology costs, training require-
ments, and many other factors will influence this deci-
sion. For data to be acquired from sources external to
the organization, the quality, cost, and timeliness of
these sources need to be considered. For example,
different market research organizations might collect
competitive sales data from retail outlets or telephone
surveys. When selecting an external data source, the
original source, the reliability of the data, the timing of
when the data are needed and when they were
collected, the precision and detail collected, and other
factors should be checked. A master data management
program often drives decisions about data sources.- Plan
- Source
- Acquire and Maintain
- Define/Describe and Inventory
- Organize and Make Accessible
- Control Quality and Integrity
- Protect and Secure
- Account for Use
- Recover/Restore and Upgrade
- Determine Retention and Dispose
- Train and Consult for Effective Use
FIGURE 4.6 Asset Management Functions