It is hard to believe that for nearly two years I have been discussing quality guru W. Edwards Deming's fourteen points of quality and their ramifications for information quality. For anyone who may have missed any of the columns, they can be found all together in Chapter 11 (The Fourteen Points of Information Quality) in my book, entitled Improving Data Warehouse and Business Information Quality, now available from John Wiley. Having just proofed the last pages the week that I am writing this column, it seems fitting to share lessons learned about information quality while writing a book on the subject.

Lesson 1. Information quality is a group effort that involves everyone in a work endeavor. Everyone is accountable for their own part of the process.

I have deep appreciation for Bob Elliott, my editor, along with Emilie, Marnie, Ellen and all the others at Wiley who went beyond the call of duty to pay excruciating attention to the details. Their guidance, suggestions and patience helped assure this was not just a book about information quality – but it was a quality book on the subject. I must also thank Diane, my wife, without whom this book would not have become a reality, and whose wisdom has inspired me. Finally, my thanks to Clay in our office who has taken on extra duties and has made excellent suggestions in his proofing and reproofing.

Everyone involved in the writing, editing, graphics layout and typesetting had specific roles. Without each performing their role correctly, quality would not happen.

Lesson 2. Information quality problems can, and will, be introduced at every hand-off in an information value chain. Minimize unnecessary interfaces.

The book publishing process has many steps, as do business processes. Each step requires specialists. But each step requires input from the previous step. One decision was made about the format of footnotes that did not get to the copy editor, and a different format was used. By the time it was discovered it was too expensive, both in time and money, to change.

Information must be shared effectively across all affected parties in a business value chain. Information sharing from common databases and sources minimizes such information "fallout."

Another example of interfaced value chain problems occurred in the graphic artwork. I use PowerPoint for my graphics, but the publisher could not accept that format. As a result we had to buy another graphics software package, Corel Draw, to redevelop the art. Costs incurred not only included the price of the software, but also the learning curve to become productive and the costs of redeveloping electronic artwork we already had. Not only were errors introduced (inadvertently), but some sophisticated features were not used initially until we learned the product more thoroughly, requiring additional rework. I had to spend much time proofing the (re)work. During the course of the year of writing, I updated a number of the graphics in the original source. These same updates had to be applied a second time to the art in Corel Draw. Furthermore, the publisher took the Corel Draw graphics and transformed them into yet another form for the actual typesetting. This introduced further errors my staff and I had to proof and correct, costing us more time – and aggravation.

The lesson is that any form of redundancy has two major problems: it increases the costs of work and decreases information quality. Businesses and information systems must cost-justify any "solution" that requires redundancy because redundancy decreases profits and shareholder value and creates information quality problems. They must further implement additional processes to control quality and correct the problems that will occur.

Lesson 3. Information quality must happen at the source – it cannot be "proofread" (or inspected) out.

Not even an army of proofreaders can find and correct defects created by the writer. When I reviewed my edited and typeset work, I found some errors that no expert editor or proofreader could be expected to find. Why? They were errors of content that only the information producer could conceivably know. Errors in footnote citations that referenced wrong pages are an example. The proofreader would need to have access to the referenced work itself, and then spend countless hours finding the correct page.

The writer or content producer is accountable for the accuracy and completeness of their work. To do so, they must have the training, incentives and accountability for their work. In cases where I had research conducted by others, they became information producers. However, as the author – and the "process owner"– I hold ultimate accountability. The same accountability applies to business process owners.

Lesson 4. Information quality is about effective communication, not about perfection or zero defects.

The reality of quality information is that it enables the information customers to perform their jobs effectively and efficiently. The quality of a book does not depend on the degree of typos and grammatical errors (I intentionally used grammatically incorrect statements – to the chagrin of the editors – to provide gender-neutral language without being awkward). The quality of a book is judged by the degree to which the reader can understand and apply the content (whether a novel or a professional book) to accomplish their objectives. To be sure, some data must be zero-defect data to meet objectives. In healthcare, for example, some inaccurate data can result in death. Information must communicate, be well-defined and presented clearly because it is much more than a simple matter of accuracy.

Lesson 5. There is a higher requirement for information quality in electronic data than in human communication.

People can read through typographical and grammatical errors and still understand the message. Computers "reading" the same data cannot "mentally" interpret what was intended without the application of sophisticated artificial intelligence techniques. Duplicate customer records, one of which has a transposition of two characters in the name, cannot be matched by a simple computer compare operation. A person looking at the two records can recognize the transposition and know they were in fact the same customer. Therefore, it is much more important to assure the correctness of electronic data if computer operations use it to make "decisions" and perform processes. Lack of attention to precision of computer data has far more potential for computer process failure than the same data "processed" by people.

Educate electronic information producers to the processes using the data they create along with the consequences of downstream process failure that results from even simple, inadvertent error.

Lesson 6. There are both "real" and "perceived" requirements for information quality.

Real requirements for a product are those required to accomplish the customer's objectives. Perceived requirements, however, relate to ancillary, subjective or aesthetic needs. For example, a book on information quality is expected to not only have good content, but be free of typos and other errors that might be "tolerated" in books on other subjects. Typographical defects might detract from the content and cause the reader to doubt the accuracy of the message. While all involved in this book have gone an extra mile to eliminate such defects, some may have been missed. I trust any such defects will not detract from the message.

The lesson is that those who talk about quality must walk it. Both management and practitioner who would exclaim the importance of information quality must practice it in action and in rewarding the behavior they expect to change.

Lesson 7. In the end, information quality will be "measured" by the end customers of the information product.

The "customers" who buy my book and apply its concepts will assess its quality. Will it help them to solve their information quality problems? If it doesn't, it lacks quality. If it does, it will be a quality book, not just a book about information quality.

Those knowledge workers who use any set of information will judge its quality. Will it help them perform their processes and solve their business problems? If not, it lacks quality. If so, it will be quality information, not just data.

What do you think? Send your comments to or through his Web site at

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access