Acknowledgments: Work reported herein has been supported, in part, by MITís Total Data Quality Management (TDQM) Research Program, MITís International Financial Services Research Center (IFSRC), Fujitsu Personal Systems, Inc. and Bull-HN.

Total Data Quality Management . . . Zero Defects Data Capture

by: Ann Marie J. McGee

Director, Strategic Marketing

Background

In the 1980's, the tools, concepts, and philosophies of managing for quality began to be known as Total Quality Management, or TQM. A company making a true commitment to be a TQM company focuses on several facets of their business including top management involvement, employee involvement, training, process identification, and supplier quality. TQM companies are committed to continuously improving the way they do business and to improve the quality of their services to their customers.

An important TQM metric is the Malcolm Baldrige National Quality Award. This award was established in the U.S. in ] 987 with the purpose of promoting quality awareness, to recognize quality achievements, and to publicize successful quality strategies. Up to two awards are given per year in each of three categories: manufacturing, service or small business. Companies apply for these awards and go through extensive examinations and site visits that address all aspects of quality improvement. The first two awards were given in 1989 to Xerox Corporation's Business Products and Systems Division and Milliken & Company. Recent recipients have included Motorola and Federal Express, with the latter being the first service company to receive the award.

Total Data Quality Management

Corporate commitments to TQM programs have resulted in stated goals that include 100% customer satisfaction. Yet 100% customer satisfaction requires that the data that is used to achieve customer satisfaction be 100% accurate, 100% free of defects and available in a timely manner. Just as Just-In-Time manufacturing allows a company to respond to the market and most importantly to the demands of individual customers, Just-In-Time-Data allows a company to respond to its customer base with a true commitment to customer satisfaction. More often than not, this is not the case.

As a result of poor data quality, many corporate automation initiatives such as sales force automation, computer integrated manufacturing, and just-in-time logistics have failed or had only partial success. A company committed to TQM programs and goals must be equally committed to information systems providing the right information at the right time in the right place or they cannot achieve 100% customer satisfaction. Managing information requires an understanding and commitment to information as a perishable asset.

Total Data Quality Management, or TDQM, addresses these data quality management issues. Making information accessible to those who need it, in addition to providing the appropriate tools for accurate and timely interpretation, are basic requirements for TDQM. Today TDQM is a basic requirement for achieving 100% customer satisfaction.

Data Quality

Data quality is more than data accuracy. Depending on the context, other dimensions may include believability, relevancy and timeliness. In order for data to be used and managed effectively, it must be credible.

The impact of using defective data represents a real hidden cost to a company's bottom line. Estimates on defects range from 1% to 5%. Even a 1% defect rate can lead to an enormous impact on the way a company does business and makes decisions, and ultimately impacts total customer satisfaction. The hidden costs associated with anything less than 100% customer satisfaction are enormous.

One of the biggest causes of defective data is field originated transaction errors. If the data is captured in the field with a 1% error rate and then keyed by a third party at a separate time (usually out of the context in which it was captured) for another 1% error rate, the data may already be 100% defective. Taking this a step further to the actual use of the data, the cost of decisions to make revisions or changes based on the data, could ultimately cost more than the actual cost of the product.

This is one of the most important issues that has resulted in moving TQM from being associated only with product quality, to being associated with total data quality. Product quality that can be corrected at the manufacturing site saves a considerable amount over product quality that must be corrected once it is in the field. Once a product reaches the field, the cost to correct a quality problem may be 10 to 100 times the actual cost of the product.

The same is true for information. The total cost of correcting a customer service problem is far higher than a product quality problem. Information is a perishable asset and must be managed as such. Capturing and managing data with a 100% commitment to zero defects will allow companies to truly deliver 100%, customer satisfaction.

Data Capture Applications

Consider Federal Express' corporate dedication to the goals outlined in their recent annual report of 100%., on-time delivery, based on 100% information accuracy, resulting in 100% customer satisfaction. As one of the leading technology innovators, Federal Express realized several years ago that their long term goals would not be achieved without a means of insuring that the data on which they were basing their business decisions and operations was defect free. Their field information system insures that data captured at the customer site, the delivery site, or in transit, is both accurate and timely. This system has provided the basis for Federal Express to deliver customer service that has set the standard for their industry. Federal Express was the first service company to win the Baldrige award, as a result of their commitment to TQM and TDQM. In addition, Federal Express has obtained a substantial market share and a formidable competitive advantage that can be directly attributed to strategic decisions based on data obtained from their information systems.

Another data capture application is currently in pilot at Sea-Land Corporation's state-of-the-art container hubs worldwide. Sea-Land has made a substantial commitment to automating the logistics of their container operations. The company understands the need for field data capture to insure that the most accurate data in terms of location, physical appearance, (damage inspections) and content is available. If the information systems show that a container is in the ship yard, yet in fact that container is not there, what does a customer do? And, what is the real impact on a customer's business? Sophisticated tracking and information systems, for logistics management, damage inspections, and customer billing all rely on TDQM.

A well-known field system was implemented by Frito-Lay several years ago. This system has grown over time and allowed data to be captured for the equal benefit of both the distribution and logistics portions of the company, as well as the retail and merchandising portions. Distribution of perishable snack foods requires timely logistics and data management in order to provide premium customer service to a wide range of retail outlets. Brand managers rely heavily on data which may include the type of retail outlet, the product and the geographic location. A defect rate as small as 1% could seriously impact not only timely distribution, but also both short and long term product and brand decisions.

Consider nuclear power plants that rely on the accurate measurement of data to provide trend information. To comply with Federal regulations, trends are monitored to insure that there are no significant increases or decreases in activity which would indicate potentially hazardous situations. The effects of defective data could easily snowball when one considers the impact of operator decisions, management decisions and the ultimate impact on safety for the general public.

MIT TDQM Consortium

FPSI has recently begun working with Dr. Stuart Madnick, a TDQM information technology professor at the Sloan School of Management at MIT. Dr. Madnick is establishing an MIT consortium between members of the MIT Information Technologies group, industry partners, and related industry-specific research programs at MIT (including the Leaders for Manufacturing, the International Financial Services Research Center, and the Center for Transportation Studies).

There are three major components of MlT's TDQM program: measurement of data quality, analysis of the impact of data quality on business, and improvement of data quality. The definitions and research issues associated with each component are summarized in the following paragraphs.

Measurement of Data Quality

Although the concept of "data quality" sounds intuitive, the definition of data quality is not. A clear definition and consensus of data quality is needed to establish a context for the use of data quality in dimensions such as accuracy, believability, relevancy and timeliness. Dr. Madnick cites an example of a major transportation company completing a comprehensive study and learning that only 68% of the shipments arrived within the desired delivery window. Due to the difficulty of integrating all the necessary data sources, this fact was not known prior to this study. In addition, 77%., of the reasons for incorrect delivery were related to erroneous data, missing data, mistrusted data, or the inappropriate use of data. Research issues will include an identification of the key dimensions of data quality, precise and meaningful definitions of each dimension, methods of measuring each dimension for raw data, and a data quality calculus (DQC) for calculating the quality of derived data.

Analysis of Data Quality Impact on Business

This component addresses the value chain relationship between data quality and successful operation of the business or conversely, the way in which poor data quality adversely impacts the business. In the transportation company example noted above, it was determined that poor data quality and usage was the cause of 77%, of the delivery misses which in turn was the major reason for an estimated loss of about $1 billion in sales, based on decreased market share. Research issues for this component include the development of a Data Quality

Value Chain Analysis (DQVCA) technique to relate data quality to key business parameters, such as sales, customer satisfaction, and profitability; and the development of an economic model of the value of correct data.

Improvement of Data Quality

This component addresses various methods for improving data quality These methods can be grouped into three interrelated categories: business redesign, data quality motivation and the use of new technologies. Business redesign attempts to simplify and streamline the operation and minimize the opportunity for data errors to occur. Data quality motivation deals with adjustments of the rewards, benefits, and perceptions to encourage more attention to improving the quality of data captured by the appropriate members of the organization. New technologies can significantly change and improve the procedures for data capture through techniques such as direct data entry by humans and direct inter-computer communications. In the same transportation company example used in the two previous examples, new procedures and technologies for more timely and direct data capture were introduced, such as portable communicating data entry devices on the moving vehicles that provide both location information (automatically) and status information (through human data entry). These changes dramatically improved data quality. Research issues will include new data entry technologies, such as "pen-based" computers and new neural network techniques for handwriting analysis, portable communicating terminals, direct connectivity among information systems, data semantics and a new paradigm for system design that includes data source tagging such as time, date and location.

Data Capture Platforms

Reducing data inaccuracy is dependent on the types of data capture systems and platforms that are available. Platforms in the past have included notebook computers, hand held terminals and pencil and paper. Pen-based computing platforms appear to offer many distinct advantages over previous platforms.

Notebooks have traditionally been too heavy, have limited battery life, are a cumbersome size for truly mobile use, require hardware and software training for the end user and lack field ruggedness. As a result, systems based on notebooks have met with limited end user acceptance and, as a result, limited success in terms of results at a corporate information system level.

Hand held terminals, while being lightweight with extended battery life, require a significant training effort, are often difficult for the end user to use due to the keyboard configuration, lack the capability of ad hoc entries, and software modification is often difficult due to the use of proprietary operating systems.

Pen-based systems are designed for field use and are available today in a variety of form factors. The most successful systems for field data capture are those optimized for size, weight, and battery life. Other features important for data capture include stylus input for intuitive use and intelligent, flexible forms. Support for data communications and additional functionality can easily be added to applications as the system requirements evolve.

Additional application benefits of pen-based systems include point of transaction data validation such as zip codes, character recognition where information that has been entered can be validated by the customer at the customer site, and signature capture to confirm customer acceptance at the time of delivery or transaction and avoid potential disputes at a later date. The use of smart forms provides an intuitive interface yet provides for changes in structure or sequence depending on previous data that is input, as well as allowing for specific additions or modifications to capture unique yet critical information.

Summary

Many potential end users for pen-based systems have installed pilots for field data capture yet have either failed to connect or lacked the tools to improve data quality at the point of data origination - data entry by the end user. In addition, systems for field data capture typically require the deployment of large numbers of units and the total requirement for capital investment can be staggering. As a result, the

majority of these projects undergo a review process at all levels of the company. TDQM provides the tools for determining an ROI for the project, measurement of the data quality and quantifying the impact on corporate-wide information systems. The emergence of pen-based technologies presents a real opportunity to translate the technology to systems for field data capture that can improve data quality and ultimately improve the way a company does business.

Fujitsu Personal Systems Offices Worldwide

Fujitsu Personal Systems, Inc.	Fujitsu Personal Systems Systems Deutschland GmbH.
5200 Patrick Henry Drive	Tel: (49) 6103/8 6018
Santa Clara, CA 95054	Fax: (49) 6103/8 80 00
Tel: (408) 982-9500
Fax: (408) 496-0609	Fujitsu Personal Systems Scandinavia
	Tel: (46) 8-552-48230
Fujitsu Personal Systems International	Fax: (46) 8-552-48290
Espace Berlioz
Parc de Sophia Antipolis	Fujitsu Personal Systems Iberia
Biot-France 06410	Tel: (34) 4/427-7606
Tel: (33) 92.94.58.58	Fax: (34) 4/427-7625
Fax: (33) 92.94.58.99
	Fujitsu Systems Business of Canada, Inc.
Fujitsu Personal Systems (U.K.) Ltd.	Box 30, 5140 Yonge Street, Suite 2000
Tel: (44) 895-430001	North York, Ontario, Canada M2N 6L7
Fax: (44) 895-430002	Tel: (416) 512-3531
	Fax: (416) 512-0344
Fujitsu Personal Systems France SARL
Tel: (33) 1.47.88.20.85
Fax: (33) 1.47.88.44.89
Fujitsu Personal Systems Italia S.r.l.
Tel: (39) 6/520 0931
Fax: (39) 6/520 0940