A recent discussion linking data quality decision making to the availability heuristic – a mental shortcut that occurs when people make judgments about the probability of events in the world by the ease with which examples come to mind – shows how data quality issues are frequently rationalised as once-off exceptions, says Gary Allemann, MD at Master Data Management.
The human brain is wired to make decisions promptly based on immediate information, and as such may frequently overlook other relevant information that is not immediately at hand. The mind extrapolates from these immediate examples to build a general case. This can be useful, but in many cases, does not accurately reflect the real scale of the problem.
If, for example, a manager is aware of two or three billing issues a day, they may regard this as inconsequential. However, in practise, these may only represent those issues that clients have escalated to the manager following repeated attempts to address them at a lower level, and may be a small subset of a much bigger problem.
Studies show that listing multiple examples requires more thought, and avoids the shortcut that may cause business to write a specific data quality problem off as an isolated incident.
If the billing manager, in this example, was to think about the total billing problems escalated to them per year (rather than per day) or was aware of the effort staff put in to addressing issues before they got to the manager, then their perception of the impact of the problem may be different.
This approach serves an additional important purpose. Once off data quality projects of the nature frequently approved by business (or sold by consultants looking to deploy manual fixes) typically provide only temporary relief, as they do not address the causes of the problem and data will revert to its natural state of chaos as discussed here.
If business people are asked to list numerous data quality issues, then the broader impact of the problem should become more apparent. In my experience, many data quality issues are related and/or interdependent.
So if the hypothetical billing manager thinks about the number of bills that are returned to sender due to faulty billing address data, or the number of billing errors that are not reported by customers who are being under billed, then the total impact of poor data quality will start to become more apparent, and the business case for a proactive approach will become clear.
A top ten list is a good starting point for a business case for a more proactive approach. The real value, however, will come when analysing the responses of more than one person or business area. This should show broad trends across the business and facilitate the business case for managing data at an enterprise level – the fundamental of a pragmatic data governance approach.
A holistic approach to data quality management may start as a single project for a key business area and in most cases will pay for itself just from this initial requirement. By setting the scene for enterprise use, users will maximise investment in both technology and process by addressing multiple business problems over time.
The human brain is wired to make decisions promptly based on immediate information, and as such may frequently overlook other relevant information that is not immediately at hand. The mind extrapolates from these immediate examples to build a general case. This can be useful, but in many cases, does not accurately reflect the real scale of the problem.
If, for example, a manager is aware of two or three billing issues a day, they may regard this as inconsequential. However, in practise, these may only represent those issues that clients have escalated to the manager following repeated attempts to address them at a lower level, and may be a small subset of a much bigger problem.
Studies show that listing multiple examples requires more thought, and avoids the shortcut that may cause business to write a specific data quality problem off as an isolated incident.
If the billing manager, in this example, was to think about the total billing problems escalated to them per year (rather than per day) or was aware of the effort staff put in to addressing issues before they got to the manager, then their perception of the impact of the problem may be different.
This approach serves an additional important purpose. Once off data quality projects of the nature frequently approved by business (or sold by consultants looking to deploy manual fixes) typically provide only temporary relief, as they do not address the causes of the problem and data will revert to its natural state of chaos as discussed here.
If business people are asked to list numerous data quality issues, then the broader impact of the problem should become more apparent. In my experience, many data quality issues are related and/or interdependent.
So if the hypothetical billing manager thinks about the number of bills that are returned to sender due to faulty billing address data, or the number of billing errors that are not reported by customers who are being under billed, then the total impact of poor data quality will start to become more apparent, and the business case for a proactive approach will become clear.
A top ten list is a good starting point for a business case for a more proactive approach. The real value, however, will come when analysing the responses of more than one person or business area. This should show broad trends across the business and facilitate the business case for managing data at an enterprise level – the fundamental of a pragmatic data governance approach.
A holistic approach to data quality management may start as a single project for a key business area and in most cases will pay for itself just from this initial requirement. By setting the scene for enterprise use, users will maximise investment in both technology and process by addressing multiple business problems over time.
