108 Part I • Information Technology
needed by other units in the company. Likewise,
because of the way data were organized, it was diffi-
cult to analyze the data residing in these application
systems. A data warehouse was created whereby
certain data from each existing system and data from
new systems that were built were extracted on a regu-
lar basis and put in the operational store. In this facility
the data were cleansed and organized for analytical
applications (e.g., by product versus by order) and
transferred to the data warehouse. Analysts thus have
a historical set of data available from each plant and
for all product lines. With the data warehouse in place,
the furniture manufacturer is beginning to employ
data mining techniques (to be discussed in Chapter 7)
to aid in areas such as analysis and forecasting.
Eventually, improvements in forecasting ability and
reductions in lost analyst time from the creation of
this data warehouse are estimated to generate a
31 percent return on investment. See the box “Birth of
a Legend” concerning data warehousing.
6.Control Quality and Integrity As with employee
certification, audits of financial records, and tests for
hazardous materials or structural defects in buildings,
quality and integrity controls must be placed on the
data resource. The concept of application independ-
ence implies that such controls must be stored as part
of the data definitions and enforced during data
capture and maintenance. In addition, periodic checks
of databases should be made as part of the audit of
financial records. As with other quality assurance
functions, these audits for data quality should be
assigned to an organization that is not directly respon-
sible for storing and managing the data.
Data quality is an especially critical issue
when data are considered a corporate asset (see thebox “Good, Clean Data”). The more data are used to
support organizational operations, the cleaner the
data should be. For example, when the data are com-
bined with a customer relationship management
(CRM) application, data quality problems can lead
to mismanaged relationships and result in lost sales.
Data are essential in enterprise resource planning
(ERP) systems, CRM, and data warehousing. The
quality of the data has a direct relationship to the
quality of the processes performed by these systems.
Data quality initiatives, like master data man-
agement,can be difficult to justify. These programs
look like overhead. So, where are the benefits? Most
often the benefits come from greater confidence in
information. Greater confidence leads to more rapid
decisions. Other benefits include less time spent
reconciling data, increased customer satisfaction
(because we deal with customers in consistent ways),
and ultimately reduced costs and increased revenues.
7.Protect and Secure The rights each manager has to
each type of data must be defined. Privileges for use of
data might include definition, retrieval, insertion,
deletion, update, and retrieval of the datum by itself or
in combination with other values. For example, a
business manager might be permitted to see the
salaries of everyone in his department but might not be
able to match names with salaries. Privileges can be
assigned to programs, databases, files, individual
records or data elements, terminals, and workstations.
Use of other equipment, data, and programs might be
limited by time of day or days of the week. The
decision on who has the right to do what with data is a
delicate balance between the need to protect the
quality and integrity of data by protecting a valuable
asset from damage or theft and the right of individualsBirth of a Legend
If your job has anything to do with data warehouses, you have heard of the tale about the correlation
between purchases of diapers and purchases of beer. The statistical oddity, duly reported in at least 200
articles, is variously attributed to Walmart, Thrifty PayLess stores, or an unidentified grocery chain.
Whichever, the retailer supposedly rearranged its shelves and sold more diapers and more beer.
Where did this tale start? It appears to have come from one Thomas Blischok, now Chief
Executive of Decisioneering Group in Scottsdale, Arizona. As Vice President of industry consulting for
NCR, he was doing a study for American Stores’s Osco Drugs in 1992 when he discovered dozens of
correlations, including one connecting beer and diapers in transactions between 5 P.M. and 7 P.M.
Blischok recounted the tale in a speech, and it became the stuff of consultants’ pitches, trade
magazine articles, and ads. But did Osco rearrange its beer or diaper shelves as a result? Nope.
[Based on Rao, 1998]