The Current State of MDM

Twenty plus years after the emergence of MDM and data governance, their relationship remains unfulfilled. Both have reached a stage of maturity. There’s lots of market chatter about why governance and master data go ‘hand in hand’. So why aren’t there more proof points of them working together?

To date, the relationship is more theoretical than practical. In theory, you should use data governance to determine the business glossary of enterprise data and the policies for creating, updating and sharing that data. Nearly all of the business rules in an MDM solution are essentially governance policies – matching rules, deduplication rules, data validation rules, and data visibility rules. What’s more, many governance products contain data cataloging functionality to scan and profile source system metadata. MDM systems are loaded from those source systems and the ingestion process would benefit from data profiling and classification. 

The reality is that the majority of MDM implementations are done without data governance, and vice versa. Only a minority of MDM and data governance clients have integrated both technologies. Most have purchased them at separate times and then integrated them at a later date. Yet, the biggest benefit would be realized by having both integrated at the beginning of their implementations. 

What gives? Industry analysts, thought leaders, and software vendors (the ones who have both products) are talking about the relationship between the two products. And while there are examples of companies using both, the majority of the time, they are implemented separately. The reality is that you can implement one without the other. MDM governance policies can be documented as project requirements and then configured in the MDM system. Data governance policies regarding data matching, viewing and sharing can be documented without actually mastering the data. The truth is that there is a gap between MDM and data governance technologies. Some companies look at the time and cost to close that gap and decide it’s not worth it. Undoubtedly, they would be better off using both, but that gap causes them to pause. 

Unfortunately, it becomes a case of penny wise, pound foolish. The costs of not integrating MDM and governance may not be obvious initially, but can derail either initiative. For example:

1. Profiling data sources before mastering. What’s the # 1 factor in shortening an MDM implementation? Profiling data sources before ingesting data. What’s the # 1 overlooked step in most MDM implementations? Yup, profiling data sources before ingesting data. It’s common sense, so why doesn’t it happen? Because they are separate tools. The result is data that is improperly prepared being much more difficult to match and master, and slowing down the implementation by weeks or months. 
2. Understanding business usage policies before implementing an MDM rule that will impact future usage. The top example of this is defining “enterprise” rules for producing a golden record, only to later discover that another business process can’t use that view for a valid business reason. However, the original record may not be preserved in MDM, meaning it is useless for that business process. 
3. MDM could be a ‘physical instantiation’ of governance. If the governance tool is implemented first, MDM could transform it from static policies and views of data sources into a living master data system to govern master data. MDM makes governance real and tangible. However, if the MDM tool can’t be configured from governance policies and business glossary definitions, then MDM doesn’t become a logical extension of governance.

Some vendors are finally heading down the path to integrate their MDM and governance products. However, the integration is arms length; it isn’t deep enough to close the gaps above. At least not yet. Another potential way to bridge that gap is with modern data architectures. A data fabric and data mesh architecture may help plug those products together more easily and share a common definition of business data entities. 

There is a good reason to implement MDM and data governance together. This isn’t just a theoretical idea. It would deliver significant business value in the real world. Modernization of MDM and governance, and overall data management architecture, may bring these two technologies together more closely to turn theory into reality. As we look to the future of MDM, this will certainly be an aspect where MDM must improve and modernize.