This is the eighth in a series of discussions of quality guru W. Edwards Deming's Fourteen Points of Quality and their ramifications on data quality. Deming's point 5, "Improve constantly and forever the system of production and service," holds two messages. The first is the recurring theme that quality must be built in at the design stage. Quality begins with intent. Management establishes intent by creating constancy of purpose for quality. Management must translate this intent into plans, specifications and tests that deliver quality to the customer.1
The second message of point 5 is that improvement is not just a one-time project; it is a continual, unending process, involving everyone in the enterprise. We each should ask ourselves every day, "What have I done this day to advance my learning and skill on this job, and what have I done to advance my education for greater satisfaction in life."2 An important byproduct of quality point 5 is that when we improve the processes for data quality, we also improve our satisfaction in work and in life!
This month I address the ramifications of quality point 5 for data quality improvement in business processes. Next month I will address its ramifications for the processes of application and data development.
Deming cites Joseph M. Juran, another leading quality consultant, for the following illustration. A hotel catches on fire. Someone yells "fire," grabs a fire extinguisher and pulls the fire alarm. The fire is extinguished. But does putting out the fire improve the hotel? No. It is not quality improvement--it is simply putting out fires. And putting out fires is not the same as process improvement.3
Data quality improvement is not the same as data cleanup. Data cleanup is fire fighting. Fixing bad or missing data is not data quality improvement--it is putting out fires. Data cleansing and transformation of data into a usable state for the data warehouse knowledge workers is fire fighting. It is the manufacturing equivalent of "scrap and rework." What then is data quality improvement?
Improvement that leads to genuine enterprise benefit is the improvement of processes that create, maintain and deliver information. Data production process improvement requires teamwork between the process owners and data producers and the downstream knowledge workers and process owners.
Data quality improvement means getting the information producers and customers together to rethink and redesign the processes where data defects are caused. Both management and professionals in the information value chain are required to improve processes. Professionals who perform the processes know them. Management must enable staff to make process improvements and to communicate that it is imperative to always look for better ways of performing the process.
Process owners cannot define quality for the data created by their process if other processes depend on that data. Only the downstream knowledge workers, or customers of that data, can define its quality and, therefore, the data specification and requirements. So to improve data quality, the process owner must look to the downstream processes that require it for data quality requirements. Then the process owner must involve the data producers to determine how to improve the process to meet downstream knowledge workers' expectations.
Data Quality Improvement Processes
1. Innovation in information products and information services. Develop information that you do not currently have. Much work may be performed because of a lack of knowledge. What is it that you do not know today, but if you did know, would fundamentally change what you could do?
2. Innovation in the process that creates data. Hand-held computers eliminated the need for utility meter readers to physically read the dials on meters. Embedded chips in the meters eliminate the need for utility meter readers. This is process innovation.
3. Improvement of existing data (data cleanup). Correcting inaccurate or missing data makes that data usable for the downstream knowledge workers.
4. Improvement of the existing process. By analyzing defects and their cause, you can identify improvements that prevent the defects from occurring. Defect prevention eliminates the costs caused by the defective data as well as the costs of correcting it to make it usable. The costs of correcting data can be 5-10 times as much as the cost of defect prevention.
All of these processes are a part of a data quality program. But none is sufficient by itself. For example, improvement to inventory management processes to reduce the cost of inventory becomes wasted effort when the competition introduces just-in-time inventory management.
Data cleanup becomes an unnecessary expense if processes can be improved to eliminate data defects. But if required data cannot be used because of unacceptable error, it must be cleaned to become usable.
Data Quality Improvement Steps
* Define the information value/cost chain for a collection of data, such as customer, order, loan, etc. This identifies all processes that can create or update the data, as well as those that use the data.
- Identify the data producer and process owner roles.
- Identify the knowledge worker and downstream process owners.
- Conduct a data quality assessment to discover the degree and nature of data errors.
- Identify categories of data defects and prioritize them based on their economic impact.
- Analyze the history to see what variables and patterns are involved in the data defects.
- Bring representatives from the pertinent processes to analyze root cause.
- Identify potential causes and agree on the most likely cause(s).
- Plan process improvements and implement them in a controlled pilot environment.
- Study the implemented improvements to see if they achieved the intended quality.
- If so, roll out the improvements and make them "permanent."
- Look for the next critical data defect to eliminate.
To conclude, data quality improvement is not a one-time activity. Data quality improvement is a cultural mind-set that says the status quo is not sufficient. It is the actions and behavior that implement the philosophy of providing quality to all our customers, whoever they may be.
What do you think? Send your comments to LEnglish@infoimpact.com or through his Web site at www.infoimpact.com.
1 Deming, Out of the Crisis, p. 49.
2 Ibid., p. 50.
3 Ibid., p. 51.
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
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access