While data quality is important to all enterprise data initiatives, some efforts only pay lip service to data quality without really making it a priority.
For example, many ETL (extract-transform-load) projects focus on performing just enough transformation on the extracted source data to be able to load it into whatever the target system is. Data profiling, if performed at all, is focused on relational and structural integrity. And that simply means "Do we have the key relationships to perform the source joins, and are the source data types compatible with the target?" Data standardization is usually limited to validating the values or formats in a few columns, and data matching is typically only performed during an initial load to remove near-exact duplicate records.
By contrast, master data management (MDM) focuses on providing the enterprise with a single, unified and consistent view of key business entities. That covers parties, products, locations and assets from common data elements across all possible data sources so a subject area can iteratively refresh and maintain its best master record. Which is why, as David Loshin explained, data quality is one of the fundamental aspects of MDM.
Data profiling is used to evaluate sources for master data entities, including performance of a baseline assessment of potential data quality issues that must be addressed. Postal validation and address verification is essential for location master data. And since most master data originates in free-form text fields (e.g., customer name, product description), the composite data elements (such as given name, family name, unit of measure, packaging type) must be parsed and standardized. Matching and survivorship processes must be used to create the best master data records. These data quality rules enable the interactive review, approval and documentation of how MDM’s single views of these key business entities are constructed. This provides both metadata lineage and data linkage back to the originating master data source systems.
Further (as Loshin explained), “because MDM programs are essentially predicated on the need for more effective sharing of trustworthy information, the aspects of MDM that encompass data quality improvement are likely to deliver business benefits that will extend beyond the funding cycle for an MDM project.” These benefits include how MDM provides well-developed processes for assessing data quality within different business contexts – and framing data quality issues within quantifiable business impacts.
Download a strategic guide to data governance and MDM