When tackling data-quality issues, the CIO might be well served by adopting a product view of information and properly regarding data as the constituent raw material from which meaningful information is manufactured. Considering data in this product-centered way enables you to leverage many tried-and-true quality principles and literally take a Total Quality Management (TQM) approach to the management of your data quality.
At the forefront of TQM are the principles of the late W. Edwards Deming. In Chapter 2 of his book Out of the Crisis, Deming expresses his quality-management philosophy with his celebrated “Fourteen Points.” In this article (and in a follow-up piece), I’ll apply Deming’s Fourteen Points to the issue of data quality. Here, I’ll examine points one through five; the remainder will be considered in the next article.
Point One: Create “constancy of purpose” toward improving your product’s quality
Constancy of purpose means getting everyone on the same page. As CIO, you must lead the charge as cheerleader, orchestrator, and motivator to drive the notion that everyone contributes to the data-quality objective. For example, while it’s good to think in terms of preventing errors at the point of data entry, don’t stop there. Why? Because the world is constantly changing. Even data that is input correctly often will not remain correct for long because the real-world entities it represents change with time. People die, or they marry and change their names, and companies are renamed or realigned through acquisition. Obtaining and maintaining an acceptable level of data quality takes the concerted effort of a motivated team supported by efficient methods and processes. Data quality is not “the other guy’s problem.”
Point Two: Adopt a quality philosophy
Your organization must also passionately embrace the goal of defect-free data, just as you’ve likely done with more tangible products. In our current economic age, information can be leveraged both to gain operational efficiencies and to sustain competitive advantage. Quality data is fundamental to these objectives. Even in simple accounting terms, poor-quality data will increase your costs, while quality data can lower them.
Point Three: Cease dependence on rework to achieve quality
I’m convinced the cost of reworking data to meet specific information requirements is quite high in most organizations. Producing the monthly sales report can require days of effort. What’s interesting is that the wasteful rework is tolerated in the context of data, while rework of tangible products is not. This is likely so because data isn’t traditionally viewed as a product. It may not even cross the minds of management that paying people to rework data is, in fact, rework. It’s somehow just accepted as being “the way it is” with data. As CIO, you can reduce rework costs by applying the principles of TQM.
Point Four: End the practice of focusing on price tag. Instead, seek to understand and minimize total cost
The cost of any item is merely an attribute and is meaningless as a measure of quality. Purchasing poor-quality lists from third parties or cutting corners on data input or data maintenance may cost heavily, all things considered. Always focus on total cost, including the cost of rework as well as the opportunity costs associated with not having the right data at the right time.
Point Five: Continuously improve your system of production to improve both quality and productivity
Here, you’re at the very core of TQM, where the notion of “continuous process improvement” causes you to focus on the processes, which produce results, rather than on the results themselves. Your goal is to design, deploy, and then continuously improve your data processes. Doing so will contribute most heavily to both quality and productivity. Understand that TQM seeks to get it right the first time and eschews the expensive “I’ll fix it later” approach.
Do flow diagrams of your current data processes, both as individual processes and as the “value-chain” of processes through which your data flows. Identify data producers (those who take down the data) and data intermediaries (those who transcribe it into another form without adding other value). (For example, the data producer takes a sales order on the paper form, and the data intermediary enters the information into the system.)
Your goal in reengineering these processes is to eliminate non-value-adding steps. This reduces both the amount of data and the number of errors. You’ll want to consider, for example, point-of-sale entry, bar coding, and other technologies aimed at getting your bits into the system both faster and closer to the origin.
Construct a feedback loop
Feedback is essential for the survival and improvement of any process. Feedback should come from “customers,” or those downstream of your process—both people and systems—who consume the data you’re producing. Make certain you have mechanisms in place to encourage, gather, report, and subsequently act upon feedback. It’s particularly important to present feedback to the producers of the data so they’ll know if they’re meeting expectations. Feedback may come from formal audits and customer satisfaction surveys completed by downstream knowledge workers. Feedback may also originate any time errors are uncovered.
Speaking of feedback, the response I received from TechRepublic members to previous articles on data quality suggests that some readers may expect a ”single pill” solution, a prescribed series of easy steps to ease the poor data-quality pain. As with tangible products, there’s no such prescription. Instead, I believe you’ll gain most through the adoption of a data-as-product perspective, a TQM philosophy (as embodied in the Fourteen Points), and the application of some creative thinking.
Finally, recognize that all of this doesn’t happen overnight. You’re dealing with paradigm shifts and deeply rooted methods and systems. Start small. Keep the faith and momentum will build. Finally, keep one of Deming’s key admonitions in mind: Quality is achieved and maintained by attending to each and every one of the Fourteen Points in your overall strategy. I’ll cover the remaining points in my next article.
Send us your data-quality tips
How have you addressed the issue of data quality in your organization? Share your tips, questions, or problems with Dan Pratte and other TechRepublic members. Send us an e-mail or post a comment below.