In today's IT world, change has become a normal part of life. New hardware and software are continually being introduced. One of the claimed benefits of these changes is improved data quality and therefore better information upon which a company can base its decisions. The addition of new hardware and software will not automatically improve the quality of your data, and unless the changes are properly incorporated into your processes, they may have a negative effect on your data and the resulting information. First it is important to remember that if a process containing problems is automated, it will still contain those problems. Automation, by itself, will not fix or improve a bad process. Problems must be resolved prior to the automation. Also upgrading your hardware or software will not fix problems. It just moves them forward and makes managers wonder why they spent money for the new hardware or software. The proper approach is to correct the problems before automating the process, moving it to new hardware, or buying new software.
THE BALLPARK
To effectively automate, or enhance the existing automation of a process, every step of the process must be evaluated and understood. At one time most processes were made up of numerous steps, some computer based, and others manually performed. Those manual steps often provided the control points. For example, a real person balanced input totals to output totals. When a person was involved in the balancing process they did more than just make sure the numbers balanced. They made sure the numbers were reasonable. It was easy for them to confirm the numbers are in the "right ballpark." They also are aware of recent experiences that should affect the numbers.
As a result of automation, many of these manual steps get combined or eliminated. While the computer can recognize that 256 records coming in does not balance with 132 records going out, can it tell you if 256 incoming records is reasonable? Obviously it is possible to program expected ranges into balancing routines, but how often is this actually done and how frequently are those expected ranges updated? When you automate the entire process, including balancing, these issues must be taken into consideration. Otherwise you give up a little bit of your quality control.
A few years ago most policy applications came in from agents and were reviewed by underwriters. These underwriters made sure the business met their company's standards. They also monitored the type and variety of business that was coming in from the agents. Now, many companies have automated the process by allowing agents, or insureds, to enter policy applications, via the Web, into automated underwriting systems. Generally, these automated systems underwrite each policy individually, based on its individual characteristics. This process may increase efficiency and reduce costs, but what does it do to your book of business? Are you now covering too many risks of a certain class or in a certain area? Did your data quality suffer due to the removal of the underwriter reviewing the application?
PAPER, PLEASE
The most common way companies combat this possible loss of quality control is to have additional reports created. These reports can be very helpful since they can easily show current totals and trends. It is also possible to have the system reconcile various individual reports and flag the situations that are outside of acceptable parameters. This can be a big step forward for data quality, if anyone looks at the reports. A few years ago most reports were on paper. Recipients at least glanced at the information prior to filing the report. Now many reports arrive electronically or are put in an electronic library of reports for inquiry. It is very easy for people to get busy and forget to examine these reports in a timely manner. Again, technological progress that should have led to an increase in quality can actually lead to a loss of quality control, as the human control points become ineffective.
Within our company we had a process in place that developed a submission data file that went to an outside rating bureau. One of the last steps in the process produced a set of paper reports that contained control totals. These totals were examined by an experienced analyst who verified that not only did they balance between input and output, but that they also balanced to other company reports and "felt right" based on that person's experience. This paper report was accompanied by the physical tape containing the submission file. If the report looked correct, the tape was taken to the mail room for shipment to the bureau.
This process was updated a couple of years ago to take advantage of modern technology. Once the submission file was created, it was immediately submitted via secure FTP. By the time the report reached the analyst for verification, the data file was already in the inbox at the bureau. If a problem with the information on the file existed it would have been too late to correct prior to submission. Fortunately, the analyst discovered this situation before any problems occurred. The system was then modified so that the analyst had to trigger a release function before the file would actually be submitted.
We all know that change happens and technological progress is a must for our companies to control expenses and stay competitive. These changes generally control expenses by eliminating manual processes. It is our responsibility as data managers to continually monitor our processes to ensure that necessary controls, sometimes including manual check points, are in place so that we can provide accurate information. This information can then be used with confidence to make decisions regarding the future of our companies and industry.
Mike Freel is a bureau statistics manager at Des Moines, Iowa-based EMC Insurance Group Inc.
(c) 2008 Insurance Networking News and SourceMedia, Inc. All Rights Reserved.
