Approximately three years ago we had a data crisis. We failed on a major software-enabled business re-engineering project, largely due to poor quality data. Ouch!

We still needed to solve the fundamental business problem, so we formed a new program to accomplish three goals: clean the data, establish processes to keep the data clean and keep both the data and the process in a state of good configuration management.

Many, if not most, organizations have data quality problems. There are a variety of methods and tools for cleaning data. Cleaning data is not usually an easy task; however, keeping data clean is an even bigger challenge. Data stewards help achieve this goal by focusing on the data process as an operational role after the cleanup project is complete.

Our biggest data challenge was the bill of material (BOM) which is a hierarchical data structure with many specific rules. Our standards had not been clear for years; consequently, we had many disparate bill structures. Additionally, our bills of material were locally ­– not globally ­– optimized, and this was a problem. In order to run global supply-chain planning tools, we required a global BOM standard and cleanup. Failure to do this correctly would be costly.

First, we created a cross- functional change board with representatives from manufacturing, engineering, planning, quality, IT and finance. This board determined the standards. At times, we approved rules one at a time. Although this may seem tedious, it actually worked well because we gained knowledge and attained teamwork when working with the simpler rules. By the time the more complicated rules and bigger issues came to the team, we were ready. In all, more than 130 rules now describe the bill-of-material standards.

We then embarked on the journey to clean all the bills of material. We soon realized this would be a slow, tedious task because we had to meet with the product engineers for each individual bill of material. With tens of thousands of bills of material and hundreds of engineers worldwide, this was a challenge. With persistence, determination and a lot of help from team members all over the world, we succeeded in cleaning all bills of material! Our challenge then became: How do we keep the bills of material clean so we don't have to do this again?

Our bill-of-material data owners are the product engineers ­– and there are hundreds of them. They know the exact content for the product and, therefore, are responsible for the content for each bill of material. We developed a specification and a process to create, modify, review and delete bills of material. We created metrics to measure the adherence of all bills of material to the standard, yet this was not enough. Although we cleaned the data and trained everyone on the new standards and process, we were not able to meet the project objective to keep the data clean. This was not acceptable. New defects slowly and regularly appeared. Further training was conducted, yet intermittent defects and other bill of material issues still appeared. We decided the correct approach was to create a bill of material data steward.

The data steward watches the data process ­– in this case, for bills of material. The data steward does not own the content; the content responsibility must remain with the data owners. The steward observes the process, notices when something goes wrong or needs attention and then takes action. The data steward sends defects back to the data owner or the person that created the defect. The data steward function is not a full-time job, but rather a role. It is part of the person's job duty, and it is not typically the only duty this person has.

The data steward is a subject-matter expert for the data specification, including where and how the data is used and all the details about the data process. The steward looks for common problems and finds ways to solve them as well as opportunities to improve the process to make it simpler, faster and better. The data steward uses the metrics to check the quality of the data and, more importantly, the quality of the data process. The best data stewards benchmark other data stewards inside and outside our company to learn and improve.

We started with one data steward. The data steward concept has since propagated to more than fifteen key data sets, and it is still growing.

Data stewards meet regularly –­ sometimes even weekly ­– with the data owners to review progress toward data quality goals and to communicate about progress, problems and ideas. These meetings are usually conducted via conferences over the intranet because the stewards and the owners are geographically dispersed.

We host virtual data-steward team meetings to share experiences and learn from each other. The data stewards have data goals, and these goals are part of their performance expectations.

Data stewards generally report to the organization that has the most to gain by keeping a data set clean. Thus, there is an intrinsic motivation for the process. Good domain knowledge, communication skills, problem-solving skills and intermediate to advanced skills with spreadsheets are the desired skills of a data steward.

The initiation of data stewards has been a key success factor in our data quality initiatives. We have found it is essential that someone owns the data process to ensure it works as designed, takes action when it doesn't, determines root causes and fixes problems so the data process is continuously being improved, measures the data quality and measures the data process. The hero that does this is the data steward!

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