Cutting Corners on MDM Comes Back to Bite IT

Loraine Lawson

I've written stories in the past about the two flavors of master data management-analytical, or and operational. These are both terms you'll hear frequently, and back in 2009, quoting a 2006 Intelligent Enterprise article, I summed up my understanding of the two thusly:

...operational MDM synchronizes master data, including product or customer information, across the company's transactional applications, while analytical MDM reconciles data "drawn from a variety of sources to deliver integrated, consistent business intelligence across the entire business."

I didn't realize that Gartner analyst Andrew White coined the term "analytical MDM," but in a recent post, he fessed up and shared what happened to one IT division that essentially opted for analytical MDM and ended up with a nightmare scenario. It's a morality tale that should give pause to those of you who might be opting to try to fudge your way through situations where master data management would be the best solution.


Actually, at first, the post is a bit confusing because he starts out explaining the concept of analytical MDM and how it differs from operational MDM. He predicates all of this on a common assumption: Analytical MDM isn't actually a real thing - "Given that we can all agree (really?) that there is no such thing, literally, as 'analytical MDM'" - and then proceeds to explain how to do it.


After several reads, I think what he's saying is that this isn't MDM, but it does seem to accomplish the same thing -- give you a single view of the data. The problem is, that's happening downstream, in the business transaction and operations applications, rather than upstream, in the applications. It's done with passive governance, through rules and transformations, rather than active governance based on policies and business users.


In other words, it's like a high-tech sleight-of-hand, making it appear the data agrees for the purpose of your BI tool, when in fact, the disparate data still exists at the sources.


This is where our MDM story begins. Enter a Frugal IT division at a global media/marketing organization. Our heroes are charged with rolling out a global business intelligence project.


This requires our heroes to synchronize the data, by the company doesn't have a full-fledge MDM project. So, our hereos do what White calls a "passive" data governance program that basically put IT in the role of data steward over the BI data warehouse.


The plan was for IT to manage the data so as to get "some semblance of 'single view' without all the cost and overhead of an operational MDM program in the business," writes White.


Do you sense a plot twist coming on? You should.


Three or four years into the project, enter the business stage left. The business is now interested in a full MDM program and wants to be actively involved in governing the data. Sounds great! Just scale the analytical MDM and yippee!


But soft-what light through yonder window breaks? It is the dawn of realization, in which IT finds out its approach doesn't scale.


The plot thickens. Writes White:

The organization was pulling their hair out since they could not figure out how to support an operational MDM program with the tools they were implementing. Bottom line- they can't. They selected tools and technology for use downstream in BI (they fell for the vendor line) and their 'analytical MDM' vendor/technology of choice is ill-suited to support a fully fledged operational MDM program.

I'm sure that IT department would love to wake up and find it's all been a bad play, put on for your amusement. Alas, it's not. The group has lots of re-work ahead, writes White.

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.


Add Comment      Leave a comment on this blog post
Jul 14, 2010 4:41 AM Andy Hayler Andy Hayler  says:

While this story is interesting, I think it important to understand that analytic MDM can have its place.  There is little doubt in my mind that in an ideal world the business would take ownership of data and embrace the governance processes required to resolve inconsistent data.  In paralle with this they would implement data quality solutions at source and put in place operational MDM solutions to provide a single version of shared master data that can be served up to the operational systems, perhaps via a service bus architecture.  However not all organisations are ready for such an approach, which nvolves considerable political will as well as resources. 

It may well be that the highest priority problem that a company faces is inability to analyse its operations across the enterprise (for example, wanting to accurately track profitability by customer, product and channel) rather than operational data issues. In such a case it makes perfect sense to begin the process with an analytic MDM approach that can deliver consistent, consolidated informatoin without the wrenching change that operational MDM brings.  In some senses you can think of this as the low-hanging fruit. 

The organisation may then choose, as a second phase, to fix their data issues at source and achieve operational efficiences.  Our own research shows that many companies who start with analytic MDM move on to a later phase of operational MDM, and in some cases vice versa.  It is true that you may need different MDM technologies to help with these phases, but to be honest few if any MDM technologies today happily deal with all data domains equally well, and both operational and analytic MDM simultaneously, so few companies at this point will end up with a completely uniform MDM technology stack across the enterprise anyway.  Much of the work done in implementing analytical MDM will not be wasted if a later operational phase is put in place.

Aug 18, 2010 10:17 AM Per Bjorkegren Per Bjorkegren  says: in response to Andy Hayler

I don't understand why it is important do discuss what is analytical and operational regarding MDM. The only way to do is to look at the complete flows from A to Z. Of course there are different application viewpoints, how to secure the right data content/quality, and how to secure deployment of master data changes..

Dec 3, 2010 2:11 AM SS SS  says:

...Reading this article made me feel glad that I no longer work as a consultant in this BI/MDM/CDI/op-MDM/anlyt-MDM/whatever.  A large chunk of my time was wasted on reading drivel from not-as-bright-as-they-think-they-are consultants constantly bringing up distinctions, acronyms where none was needed, always promising that the sky would fall down unless you did this one thing that I am talking about. And even that would have been fine, except that even that "one thing" was never cleanly explained other than in MBA/BS-speake


Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.