Data Centers

Use these Fourteen Points to improve the quality of your data

In today's business climate, poor-quality data can threaten the livelihood of your enterprise. This article shows how you can apply a quality-management expert?s Fourteen Point program to improve your data ?product.?

More than ever, enterprises are dependent on data quality because it is our data—and the information we craft from that data—that differentiates a company in the competitive landscape.

It is my belief that a CIO’s quest for high-quality data should begin by adopting a “data-as-product” view. Taking this perspective on data allows you to then enhance the quality of your data by applying the time-tested quality-management concepts of W. Edwards Deming. In Chapter 2 of his book Out of the Crisis, Deming summarizes his quality-management philosophy with his celebrated “Fourteen Points.” In this article, I’ll apply Points Six through Fourteen of Deming’s Fourteen Points to the issue of data quality. (Click here to read my article explaining Points One through Five.)

Third in a three-part series on data quality
This article concludes Dan Pratte’s trilogy on the problems and costs associated with data quality in the modern enterprise. In the first installment, he discussed the importance of data quality and how data warehouses and marts can exacerbate any quality problems. The second article in the series began Pratte’s coverage of W. Edwards Deming’s Fourteen Points and their application to the issue of data quality.

Point Six: Institute training
It’s okay to assign responsibility for quality to those performing data-entry work, but individuals simply must be trained—both in proper methods and in data-quality standards. According to Deming, accountability without adequate training can only create fear and apprehension.

I was once involved in a project where one person’s data entry always appeared in uppercase, requiring downstream rework. When the problem was brought up, this individual’s response was something like: “No one gave me instructions when I started here” and “They just showed me the desk and the computer, and I went to work.”

If you expect data quality, train your data producers properly.

Point Seven: Institute leadership. Aim to help people and machines do a better job
Leaders are coaches, not cops. Great managers are leaders who encourage improvement and create an environment conducive to enterprise quality and productivity goals. Leaders play a crucial role in improving data quality by helping their workers solve problems, by implementing better machines, and by always looking for a better way to accomplish goals.

Deming also advised, "Leaders must know the work that they supervise." Without good knowledge of the process, management can’t effectively coach, and managers end up focusing on quantity, not quality.

Point Eight: Drive out fear so that everyone may work effectively
In many work environments, employees are afraid to ask questions, so they continue to do things the wrong way. "The economic loss from fear is appalling," Deming said. If people feel secure, then they will interact, be more productive, and produce a better-quality product. You must take very strong steps to encourage honest, open dialog.

It’s also important that data defects, once discovered, are used not to “punish” data producers, but to identify and correct defective processes.

Point Nine: Break down interdepartmental barriers
Data you collect through your various business processes becomes the “raw material” for other business processes. Some of these processes are familiar—billing and payroll for example—and some are less so. For example, information itself is both the product of a process and the input for the decision-making process.

Any movement of data through the enterprise in this way, while adding value with each step, is collectively termed the Data Value Chain. In real life, the chain will almost always cross organizational and even company boundaries. Because boundaries are often barriers, Deming encouraged the formation of teams comprised of people from different areas. The team then shares the responsibility for data quality.

Point Ten: Eliminate slogans and exhortations asking for zero defects and higher productivity levels
“Quality” initiatives have left a foul taste in the mouths of many, often becoming especially tedious for workers. “We spent a lot of time counting paperclips,” someone once said. For management, Deming’s quality principles have sometimes been degraded into a series of meaningless charts and slogans. In spite of these misfires, you don’t want to confuse bad practice with good theory. Deming’s teachings are real, and they work.

According to Deming, the problem with charts and slogans is that they are directed at the wrong people, because most causes of low quality and low productivity are process-related, not people-related. So, the power to effect improvement lies beyond the workforce. Workers can do little to change the system—that lies with management.

Point Eleven: Eliminate work standards (quotas) and other objectives
One of the key problems with trying to manage by way of quotas and objectives is that the worker can often achieve them by compromising quality. Deming maintains that enforcing quotas and other work standards interferes with quality perhaps more than any other working condition. Again, the answer lies in the process that produces the product. Focus on the process, and the productivity will follow.

Point Twelve: Remove barriers that rob the worker of their right to pride of workmanship
People who feel good about their output contribute substantially to both quality and productivity, yet barriers will often prevent this. What might some of these barriers be? Barriers include conflicting or unclear goals, arbitrary decisions by supervisors, fear of mistakes or failure, insufficient information and training, and little constructive feedback on performance.

Consider it a key management responsibility to target and systematically eliminate these barriers.

Point Thirteen: Institute a vigorous program of cross-training and self-improvement
Cross-training can benefit productivity and data quality in at least four key ways. First, data producers who receive training in tasks downstream of their own gain beneficial perspectives. Second, satisfaction in the work environment is enhanced by cross-training when it’s viewed as an opportunity for self-improvement. The third involves the principle of stewardship: With a better firsthand knowledge of my data customers, their requirements, and the issues they face, I’ll seek to meet their requirements in better and faster ways. Fourth, cross-training enhances the worker’s ability to recognize when systems are drifting out of control and to take appropriate action to bring them back on track.

Point Fourteen: Create an appropriate structure
Top management must create the structure that emphasizes all of Deming’s Fourteen Points. Without this structure, no viable long-term benefits can be achieved.

According to Deming, the first milestone is passed when most employees understand the Fourteen Points and become active participants in the transformation from the old paradigm to the new one. Everyone may then become an active part of the process.

Conclusion: The benefits of quality data
Certainly, it’s clear that the quality of your data is fundamental to the quality of your information product. This is important insofar as the exploitation of information is what is driving competitive advantage in the modern enterprise. As data itself is the product of enterprise processes, its quality may be improved in the same way as traditional products—by way of Total Quality Management (TQM) tools and technologies.

I think you’ll find much of value here, but you’ll have to ponder these Fourteen Points carefully and apply them creatively. Consider them as a framework and use them to shape your thinking. Finally, don’t lose patience. Rome, says the proverb, wasn’t built in a day. Over time, taking the small steps to implement these time-tested principles will combine to deliver large gains with respect to the quality of your enterprise data.

Share your data-quality tips
Have you tried implementing Deming’s Fourteen Points? Do you have experience with other quality programs, either successful or not? Send us an e-mail or post a comment below and share your thoughts.


Editor's Picks

Free Newsletters, In your Inbox