Blog post

We are only Half Pregnant with Master Data Management

By Andrew White | April 17, 2013 | 2 Comments

MDMMaster Data ManagementInformation GovernanceEnterprise Information Management (EIM)

We must be half pregnant with MDM!  I see an increase in the number of inquiries with end user organizations that are making some pretty simple mistakes.  But upon reflection, these mistakes are mostly due to ignorance and immaturity of the IT market.  And some are due to vendors’ positions being out of line with neutral view for what the market should be doing.

There are a growing number of end user organizations that are trying to “master” as much data as they can in their MDM hubs.  In some cases, the organization might be trying to figure out how to store Bill of Material on its product information hub.  Another might be trying to store lifetime value (in terms of forecasted revenue) as an analytic within the customer master data hub.  What are we doing, people??????  This is near madness.  Just because a vendor or architect says we can, or could, does not mean we should.  Also, just because it seems like a good idea at the time, don’t just “do it” without thinking through what you put into this position (and why such a decision could perpetuate the problem you are trying to solve).

We need to spend a few minutes thinking about how and why MDM came about.  One major reason that MDM came about is because organizations did not have a repeatable, defined, discrete discipline for assuring “single version” of important data spanning most important business systems that drive what they do as a business.  Some, not all, of these organizations had massive investments in ERP systems.  ERP, among other things, could be a large suite of business applications.  As such ERP represents a very large data model – lots of master data and lots of other data used by each application and/or group of applications in the ERP suite.

But if there were other applications in the landscape, even that ERP data model was not homogenous – there were others – so the landscape remained heterogeneous. And it would remain as such even if those applications were sold by the same “ERP vendor” – for most of the larger vendors, that is, with large application suites each with different data models.

So MDM was conceived just a few years ago.  But think about – leading industry and technology “standards” bodies did not even have MDM in their models and practices and lexicons 7 years ago.  Yet now they do.  Books on “Business Metadata”, written just a few years ago, confuse metadata and master data.  So even if MDM is included, its not always properly understood or represented.  Better, leaders and innovators that were early with MDM have now figured out that MDM really only focuses on master data, and that other “layers” of the information infrastructure need to help govern that data that sits between the extremes:

  • Data widely reused, defines what we do as a business – e.g. master data
  • Data used by only one application, so thus governed by that one application

And this is the half pregnant idea I started with.  We, as an industry, have NOT yet formalized, or named, or defined, HOW we plan to govern that data that is used in many places, that is not master data (does not define the terms the business uses to say what we do as an organization) but behaves like master data.  This is in fact the larger amount of data in our organizations – much even unstructured. 

We can’t assume that we can govern this data through Content Management (CM) or Information Lifecycle Management (ILM) or Records Management or even Data Quality.  These are themselves silod practices that do not provide for an integrated governance framework, that build on MDM and expand the program to the necessary data needing governance.  Yet CM, ILM, DQ and so on are integral elements of what we do.  So something has to link this investments; something has to change those practices so they align and work in a synergistic fashion.  Just “adding MDM” is not it.

We (as an industry) don’t yet have all the answer – yet every day more and more organizations are getting started with MDM, and beginning to think about this next phase, this next stage.  We at Gartner do have an idea, using the name Enterprise Information Management (EIM) but the bad news there is that vendors have abused this name already.  Worse, some end users have too.  Some organizations just renamed their DI/DQ teams to “EIM teams” as if this would yield a real information management strategy.  Ooops.

So we have a LOT more work ahead of us.  MDM is good – it is a great start.  And EIM is a great target, an ideal end-state.  But what you do next, the explicit next step to move beyond the successful information governance program that is MDM that is the question.   We will be busy with the mid-wives for some time to come…

Comments are closed


  • Manish Sood says:

    Your observation about being half pregnant with MDM is spot on.
    However, the problem does not lie with the customers, since they are trying to solve their tactical problems with a strategic capability such as MDM. The problem lies with the software vendors, who after a decade of selling solutions for the MDM market have barely moved the needle on how to best evolve the concept of MDM. For example, it is surprising to see that most of the products in the market are still wedded to the notion of a Single View, when customers have repeatedly expressed the need for a holistic view that allows the consumer of the information to drive the definition of the view at the time or place of consumption of this information.
    Customers are putting analytic information into a MDM hub because goal they are trying to get to is operational access to unified and streamlined “Information”. This is not because they do not understand the concept of dimensional vs fact data; but this is because the vendors have not been able to answer their question of how to make this information streamlined and available as a combination of Master and Transactional data.
    This is an areas where MDM has to evolve to the next step where it bridges the gap and acts as a foundation that enables delivery of streamlined Information and allows the users of the data to drive the view that they need from the superset of the information available. This is where MDM and Big Data converge to deliver operational information. So far in our work at Reltio, we have been referring to it as MDM 2.0; maybe you can recommend a better term for the next generation of MDM…

  • Andrew White says:

    Well said Manish, and thanks for the comments. Yes, the maturity of technology remains variable. And the majority of them are still very “new” even though the discrete topic of MDM is only a few years old.

    Some vendor claim “multidomain” MDM and while it is physically possible to “model your grandmother” (as I like to say) with many MDM solutions, having real world experience with same (multidomain, not the grandmother) is very different.

    On a more serious point, the link between metadata management tools, glossaries, BPMS, business rules engines, semantic technologies and now also information stewardship applications, remains varied.

    We are all swimming upstream against 30 years of building and deploying vertically build applications (business process with analytics and data) and data warehouse. The good news is that MDM is rocking the boat, one step at a time. Onward and upward!