BI Concept: Master Data Management (MDM)

When I first introduced readers of this blog to business intelligence I claimed that, by definition, BI had to be "integrated and coordinated" in order to be successful. I gave the example of the boss who, confronted with two or more sets of sales data, wondered in frustration which was correct. I could just as easily have made the example about the boss wondering which of several customer or product lists was correct. I made further allusions to this concept in discussing capture of data as the first step in the BI data life-cycle. And this is a major consideration in using extraction, transformation, and loading (ETL) tools to clean up such problems.

Unfortunately, this cleanup is not always fed back to the front-line systems where the data are initially captured. And there may be many such systems: lead generation systems, sales order systems, accounts receivable, and so on. And in most cases front-line workers are looking at the transaction systems they use daily, not the data warehouse.

This problem is the impetus for master data management (MDM). Simply put, the goal of MDM is to ensure that the view of company data is uniform across all business units and systems. There are two pieces to this. The first piece is to create a master data set for each of the critical dimensions (such as Customer or Product) of interest. The idea is that by referring to the master data set all parts of the business have a unified view of business data. This is depicted in the slide below.


This sounds like a terrific idea, but it leaves your front-line people looking in two different places, their data entry system, and the master data, for information they need. So the second, and much more difficult, piece is to integrate the master data back into the source systems, for example as in the next slide.


There are several approaches to doing this: by programming the source system to always look at the master data, or by actually re-writing source system data with master data, or any number of other strategies.

Sounds like a lot of work!

The notion of trying to tie every major element of interest together across all business units and systems can sound daunting. Many businesses choose to focus on specific areas of the business that they feel are most critical to their success. Thus you'll hear about Customer Data Integration (CDI), or Product Data Integration (PDI), or variations of such acronyms. Think of these as subsets of MDM that are focused on those specific subject areas.

Who are the players?

As you might imagine, many of the major vendors in BI in general are well represented in the MDM field. These include IBM, Oracle, Informatica, and SAP. But there are some lesser known companies in Gartner's "Magic Quadrant," among them Tibco, DataFlux, and VisionWare.