DataFlux

Going back to my earlier blogs on shared business vocabulary for common data names and data definitions I suggested then that master data should be defined using these enterprise wide data names and data definitions. Also if you plan on building an MDM system yourself you need to build a data model for master data entities using master data definitions. Having done this, the hardest job of all is then discovering what master data is out there, i.e. what systems it resides in. This data discovery exercise is sometimes called data exploration, the objective being to identify master data in all the disparate systems so that you can map this data to common master data definitions for that specific entity.

So, for example, if you are looking for customer data it may reside in CRM systems such as a Sales Force Automation system, a customer service system, a marketing system, a billing system and a distribution system. In my experience, if you follow processes associated with customer data you start to discover master data residing in other unexpected places, e.g. Access databases, Excel spreadsheets etc. This is because it can be the case that business users record more data than core systems support.

In fact, just recently I discovered this in an insurance company for master data on insured risks. They entered some information into the underwriting system, but because that system did not permit the capture of all data, business users siply recorded the rest in Access. So turn over every stone when looking for master data subsets in disparate systems.

Having discovered disparate master data, you will most likely find that even if copies of master data reside in different systems the data may be named differently across those systems. So, the next challenge is for each master data entity (e.g. customer, product, employee, asset etc.) to map disparate master data in identified systems to commonly defined master data described using the shared business vocabulary. This is a difficult and time-consuming exercise to do manually. Fortunately, there are tools now available to help explore systems looking for master data and to automatically map that disparate data to commonly defined enterprise wide definitions of master data. Once this is done, you will understand all the different names for the same master data around the enterprise and how these map to a common definition. This, for me, is the point at which you start to conquer this master data problem because now you have the master data items commonly defined for an entity, you have identified where that data resides in the enterprise and you have mapped it. At this point you are then able to move to the next stage…. something I’ll talk about in my next blog.