Data quality expert Tom Redman recently published a White Paper on data quality that lists the 10 habits practiced by companies with the best data.
While Redmond, president of Navesink Consulting Group, has spoken about these habits before, data quality remains a big issue for most companies. Indeed, Gartner last year said that more than 25 percent of critical data in Fortune 1,000 companies was inaccurate, incomplete or duplicated and added its research found that 75% of large enterprises would make little to no progress toward improving their data quality for years.
But, as Redman maintains, companies don’t have to suffer from bad data.
The latest listing of the 10 habits included in the just-released paper “Data Warehouses and Quality: Not Just for IT Anymore,” which was sponsored by Teradata and available on the data warehouse vendor’s Web site. The 10 habits don’t necessitate some radically rethink of IT strategy, but rather, he says, just “good scientific practice to data management.”
The 10 habits include:
A focus on the customer. “It is essential because quality is in the eyes of the customer, and without understanding customer needs, it is impossible to satisfy them,” he says.
Good process management. Often times, the people that input data into a system have no idea what the data is or why it needs to be entered into a computer system. But the companies with the best data are the ones that explain why data in being collected and how it’s going to be used. Once people understand the importance of data, he says, immediate improvements often result.
Good supplier management. Companies with good data work closely with the companies in their supply chain that provide them with data. “It seems odd that companies will issue detailed specifications and employ service level agreements for the physical goods they purchase and completely ignore the data. But, the methods of supplier management for physical goods are just as effective for data.”
Measurements. The adage “you can’t manage what you don’t measure” applies to data quality as well as anything else. Simply tracking how often customers point out a mistake on a bill can be a starting point.
Continuous improvement. Companies with good data practices investigate where errors originate and identify their root causes. Then they change their processes accordingly. Most often, he says, the root causes are a lack of requirements, poorly defined processes, a lack of training, unfathomable applications (such as a call center application that doesn’t properly instruct a user on the next step in a process), and illegible inputs.
Controls. Companies with the best data employ controls at many levels, he says. They help people enter data correctly, they prevent errors from perpetuating, they ensure that systems work properly, and they distinguish between special and common causes or process variation. “The controls need to be so-called statistical controls, and they are indeed powerful,” he says.
Improvement targets. Enterprise with good data have managers that create baselines and set aggressively improvement targets – such as halving data error rates each year.
Management accountability. Those who create the data are responsible for its quality and it’s up to the direct supervisor to make sure good data handling habits are practiced. “[I]n many organizations, it is important that managerial accountabilities for data be formalized via a data policy,” Redman says.
Management of the soft issues. Companies with the best data know that all organizations are political and that people want to protect their turf. For instance, the act of getting a salesperson to input complete and accurate data into a customer relationship management system, can be a hurdle because salespeople know as well as anyone else that knowledge is power. Some companies deal with it through influence. Others many take a more draconian measure, such as holding up a commission check until a salesperson correctly inputs all the information required of him or her – although Redman doesn’t recommend this approach.
Senior leadership. Few things happen in an organization without management support. And data quality is certainly one of those things. “Those with the best data recognize this and secure broad, senior support. The higher and broader the better,” he says.
Redman states that he’s amazed by how tolerant companies are of having flawed data in their systems and ignorant of the high cost they pay for these flaws.
Companies can go out and pay big bucks for tools that check data, or send teams of people into the organization to correct bad data. But, he says, “the habits are easy compared to the alternative and pay big dividends compared to the alternative.”
Only registered users can write comments.
Please login or register.