With recent events such as HM Revenue and Customs incorrectly issuing 10,000 company fines, and retailer Argos pricing a television for GBP0.49 instead of GBP499 online, questions about these organizations’ approaches to data quality (DQ) have been raised. Interestingly, as firms deal with an ever-increasing number of data sources and more complex business models, DQ issues will also increase.
There are three fundamental issues that must be addressed to begin to tackle the problem of poor DQ. The first issue is a matter of continuing frustration for many IT managers, and concerns ownership and responsibility for data within a company. A commonly-held belief within the business community is that the IT department is the custodian of data, because most corporate data sits on computer systems. However, most financial transactions and accounting ledgers also sit on computer systems, but are ultimately the responsibility of the finance director; why, then, is corporate data any different?
DQ should be a strategic concern for the business, and can often necessitate that a senior executive be appointed to the role of data governor (DG). Primarily this role is to cross-functionally ensure that each division uses the same language and definition.
For example, sales in channel ‘A’ report profits up 15%, while channel ‘B’ remains flat. This could be a true reflection of business, or they could be using the same word (profit) but meaning different things (without overheads, and including overheads). Thus the DG’s role is to ensure that this data, when compared, is meaningful and not misleading. To perform this function they must be given the resources and remit to define the corporate language, set the company’s reporting standards, and actively manage the company’s data requirements as the business changes.
The second issue is to develop metrics for data quality, which requires a definition of what is meant by data quality. This will be different from company to company, but a generic statement, such as ‘fitness for use, which means being free from defects, encompassing all the desired characteristics, and being consistent with what the user expects’ encapsulates the fundamentals. Once you have the definition, you can then design the metrics and the associated processes to ensure compliance.
The third issue is one for the IT department, as it must provide the technology to support this change to make DQ strategic. The current approach is to close the loop between data warehouses and operational systems by using data profiling to identify problems and their root causes. Data cleansing tools can then be employed to ensure that all data has been corrected. This is not a one-off exercise, and must be built in to the business processes discussed above.
The ultimate solution for DQ being advocated currently is to implement a business intelligence (BI) and data stewardship ‘center of excellence’. The excellence center is a formal group that spans the organization, has both staff and technology, and aims to manage the corporation’s BI or data assets in order to ensure high reusability, accessibility, and quality.
Source: OpinionWire by Butler Group (www.butlergroup.com)