You can boil a successful Master Data Management (MDM) initiative down to these five essential elements:
(1) a Hub of some type – there are three major types of MDM hubs. In a Persistent Hub, all of the critical data is copied from the source system into the hub. In a Registry Hub, only the identifying information and key identifiers are copied into the hub. In a Hybrid Hub, a mix of both styles is used, giving you the ability to fine-tune how much transactional data is copied into the hub.
(2) some kind of data integration or middleware – because it’s important to be able to dynamically synchronize data into and out of the central hub. The synchronization doesn’t have to be real-time, although a lot of organizations are heading that way, or at least to “near real-time”. But since the whole point of these types of projects is to build a “Single Source of Truth” on a particular domain like customers or products, having out of date information in the hub, or even worse, not synchronizing the data quality improvements you make back to the original source system, can defeat the whole purpose of the project.
(3) data quality capabilities – most companies quickly realize (through data profiling, which is strongly encouraged) or just through looking at the data manually, that their information starts off with a fairly low level of data quality. So some type of data quality tool can be very helpful in standardizing information (changing all occurrences of a state value of “Massachusetts” to “MA”), correcting information (when someone spells “Xerox” as “Zerox”), and filling in missing information (when someone doesn’t provide a value for state or province). A robust data quality tool can make the difference between a failed project and a successful project.
(4) external content – also known as enrichment. Having formerly worked for D&B, one of the leading providers of information on businesses, I consistently saw the value of providing information you didn’t already have. It could be something as straightforward as SIC codes, or as complex as corporate family trees and credit ratings. But when you don’t know what you don’t know, having an external content provider can be a big help.
(5) data governance – I put this last, but it’s actually the most important of the five. Without the people and processes that you’ll need to develop around your central hub, the technology is (at best) going to be “a solution in search of a problem”. The business won’t embrace the solution unless they’re driving it, and resolving difficult questions of data ownership or correctness is going to take some kind of cross-functional data governance group in your business, with an executive sponsor, business data stewards, IT support, etc.
Bringing together the previous four elements of an MDM hub is difficult enough – don’t try to do it without the support of a data governance organization to own the solution as it is developed and deployed.
I’ll write in an upcoming post about the holistic approach I recommend to deploying an MDM solution, paying attention to People, Process, Technology and Information in a balanced way. But it reinforces a lot of these basic points.