The leading data profiling technologies are helpful in terms of finding flawed data, but they do very little to improve the processes and procedures that led to those flaws in the first place, according to information quality guru Larry English.
English -- the founder of Information Impact International Inc. and an internationally-recognized speaker, educator, author and consultant -- says the key to improving information quality and related processes centers on being proactive and determining the root cause of problems.
In this exclusive SearchDataManagement.com interview, English explains how quality principles from the world of manufacturing have been applied to information and talks about the difference between 'data' and 'information.' He also provides several tips for organizations that want to improve the quality of information being delivered to knowledge workers. Here are some excerpts from that conversation:
Is it true that the most popular approaches to information quality stem from the manufacturing sector?
Larry English: Here's the deal: There are proven quality management principles that have been applied to quality in manufacturing. Those principles [have] to do with the quality of tangible products and services, [but they have also been applied] to information processes. The only difference [with regard to information processes] is that the 'product' is delivered information. Therefore, we have to focus on designing quality into the processes that create and maintain and deliver information to specific knowledge workers who depend on that information. If we have errors in the information, it will cause alienation to both [internal and external] customers.
Why do you think that data profiling tools do not add value to information quality processes?
English: You have the big software makers who are making data cleansing software, data profiling software, assessment software and all of this does not add value to the process.It is true that you need to find where you have defective data. But it's not just the data in the databases, [you also have to look for defects in] the way that information is presented to knowledge workers. It's vitally important when we seek to improve [such processes] that we understand the root cause [of problems and] not just the precipitating causes. The problem with the data profiling tools is they don't tell you what the root cause is. There is no way to analyze that within their software.
The phrases 'data quality' and 'information quality' are frequently used interchangeably, but you make a clear distinction between the two. Could you explain the difference?
English: Information is the finished product. Data is the raw material. Generally, it is data that we capture and put in the database. But most knowledge workers will not have access to that, and if they did they probably would not be able to understand it just by extracting data values. Information, meanwhile, will generally be combined and presented to the knowledge workers in a way that meets their needs to be able to perform their work effectively.
How can organizations begin the task of improving information quality and information delivery processes?
English: You have to look at, on a priority basis, the most important sets of information that you have. [Look at the information that, if flawed, can cause a major failure.] For example, in the BP oil spill, the team that was working on the Deep Water Horizon rig made some changes to safety procedures, and those changes to the safety procedures were not implemented according to their process of making changes, nor did they have an opportunity to test the new procedure. As a result, there were errors and defects in that procedure that caused the explosion. [It's important to get] that information right and to be sure that if you make changes to processes that you have the ability to test them and ensure that they [work].
What else can organizations do to improve information processes?
English: To improve information processes, we use a proven technique that is called Plan, Do, Check, Study and Act. You put the process in a controlled mechanism so that you can execute that process without endangering the current production information that you have, and if it works in that mode then we can come back and roll that out [to the] official production processes. [You] then monitor them to ensure that they stay in control. Statistical quality control charts are a key mechanism for identifying if there is variation in the production -- that is, if there are defects in the information being captured that will subsequently cause other processes to fail.
Are people generally to blame for information quality problems?
English: There is no blame in quality management. People are not the cause of defects. If they are the cause of defects, it's a precipitating cause that is caused by an earlier problem which is the root cause. [The root cause] will probably be failure to provide the right resources and training for a person to do their job effectively. If there is a lack of following procedures then you have to look and say, 'Why was there a lack of following procedures?' The Japanese would ask why five times to get back to the root cause.
About the speaker: Larry English made a career of applying well-known quality systems to the world of data and information management. For example, he developed the Total Information Quality Managementmethodology applying Kaizen quality principlesto information quality management. His company, Information Impact International, provides consulting and education for information quality and related disciplines. English also chairs information quality conferences around the world and co-founded the International Association for Information and Data Quality (IAIDQ).
Mark Brunelli is the News Editor for SearchDataManagement.com. Follow him on Twitter @Brunola88.