Blog post

The Evolution of MDM

By Malcolm Hawker | February 17, 2020 | 11 Comments

Hello and thank you for visiting my blog.  As a new analyst with Gartner, I’m excited to use this platform to solicit your input on all things related to MDM and Data Governance. This blog is a perfect vehicle for you to interact with me, and your peers, on topics that might otherwise not be covered in our formal Gartner research. This is also a great way to ‘seed’ important issues that potentially should garner additional Gartner coverage.

With that thought in mind, my inaugural blog is focused on the evolutionary forces that are having disruptive impacts on the markets for MDM software and several adjacent markets.  MDM platforms are embracing a broader set of technologies around matching and data governance that are increasingly blurring the lines that distinguish MDM software from other platforms –  especially CDP’s.  (I touched on this in a recent research note about the similarities and differences between MDMs and CDPs for C360).   The biggest shifts in the MDM market include, but aren’t limited to:

– The growing use of graph and AI/ML technologies to assist in the automation of data quality and governance processes used by MDM software such as profiling, cataloging, lineage/process mapping, and relationship mapping, where these processes are increasingly run against large, non-relational data stores.
– The increasing use of graph in big data environments for entity resolution, where the strength of relationships between nodes are a proxy for the match confidence levels generated by MDM’s through compute-intense probabilistic algorithms. The rapid growth of data volumes that MDM’s are expected to master is pushing MDM vendors towards these more flexible matching paradigms.
– The idea that a gold master record could be created at ‘run time’, where that master record may not even technically be even persisted in a centralized hub – yet still be used in multiple applications and fit the definition of a true master.
– The explosive growth of MDM deployments in the cloud, where the historical limitations imposed by on-prem hardware on the amount of data that could be matched in a hub are evaporating. The move to the cloud is certainly not a new phenomenon, but what’s interesting here are the additional features / services could eventually be developed by MDM vendors who are running cloud-native multi-tenant data platforms.  ‘Governance as a Service’ and several other such services come to mind.

If entity resolution, hierarchy/relationship management, and the generation of gold master records are the ‘foundation’ of the MDM house, then it’s a foundation that is increasingly shared with other technologies.  This raises interesting questions about the nature of the MDM market – including the potential convergence of MDM platforms with other platforms and increasing competitive pressure on MDM vendors.  Thanks to the push to the cloud, these convergences could even extend outside the software world and into areas such as master or reference data as a service.

What do you think?  I am eager to hear your thoughts.

The Convergence of MDM and Adjacent Technologies




Leave a Comment


  • Great blog, very insightful!
    Wouldn’t you say that ML can augment or enhance the data management process, not just assist. It may just be choice if words, but there is an important distinction between supporting existing methodology in a better way with AI/ML versus augmenting and sometimes redefined certain MDM processes with AI/ML.

    In the evolution of MDM, what role do you see MDM playing in powering CX directly? Do you see MDM systems surfacing C360 in transactional systems like a Contact Center or Sales Automation system? Or MDM systems being able to capture insights and supporting next-best-actions or other decision engines?

    We are living in exciting times! Once again, thanks for sharing your thoughts!

    • Malcolm Hawker says:

      Venki – Words matter – and ‘augment’ would certainly have been a more appropriate choice here than ‘assist’. My point here is that AI/ML are allowing for fundamentally new methods of data governance, which go well beyond existing approaches. So thank you for catching this. To your other questions about CX – I certainly do see MDMs powering both the CX and other transactional systems – at least insofar as providing the master customer data needed to deliver exceptional experiences. Decision engines, marketing automation platforms, ad servers, etc. would consume that data.

  • Malcolm,
    Great blog and I am looking forward to your analysis of the evolution of the MDM market. We are also seeing many customers using master data as the core of the customer 360 used for CX (both analytics and real-time personalization). Those customers are also using MDM-like functions (identity resolution, survivorship rules, data stewardship) used for different purposes (confidence-based matching, creation of unique profiles for activation, making minimal match-pair decisions to train machine-learning matching engines). More marketers and analytics users are interested in using those functions to synthesize customer 360 profiles from marketing & enterprise data, to create contextual profiles & insights that power CX processes. 2nd generation CDP requirements have a lot in common with MDM. It is exciting to think of the new use cases that MDM will address as it is repackaged for new business functions & users. We look forward to continued discussions with you on this important topic.

  • Buvana Radhakrishnan says:

    The need for scalable knowledge framework will be one of the key drivers . Institutional knowledge and domain knowledge should be captured in this framework- Knowledge as service is one of the paradigm shifts that I envision

  • Steve Jones says:

    First of all “Hallelujah”, I completely agree that the MDM market is going to go through an evolution. The use of transactional records to do matching using ML provides a completely different way of uniquely identifying the “things” over traditional “golden record” approaches and has the advantage that it drives a more effective x-ref in that it works from the data volume by default.

    Its not the only answer but in a big data and data world you need your matching to leverage all the data not just the minimal sets that an MDM tool can support.

  • Excellent blog Malcolm and I agree with everything you are saying. One thing that surprises me, knowing the volumes of money software vendors are putting into R&D, that a lot of what you mention has not already been incorporated completely into some of the OOTB solutions out in the market.

    I feel these innovations have affected how the “magic quadrant” has seen such fluctuations recently within MDM vendors. I know within my organization we are innovatively taking technologies like this and applying them, but they do not always “integrate” well with a good portion of the existing versions of MDM software out there, this allowing for that governance piece you’re speaking of that ultimately prides a “complete solution“.

    • Malcolm Hawker says:

      There are some MDM and CDP vendors that would disagree with your first comment but your point is well taken – the bridge between these two worlds has not been widely established – yet. This will most certainly change given the mounting demand for these more ‘holistic’ solutions. I’m particularly interested to see how vendors like and Adobe will respond. Will these solutions embrace more MDM-like functionality? The recent launch of Customer 360 by SFDC definitely appears to be a step in that direction.

  • narmafzar says:

    thank you for sharing such nice and useful information
    A second area of evolution would be around new emerging technologies,” remarked Bhatia. The companies that are looking to the future are the ones that are making the right investments and asking the questions such as: “I’ve got all this IoT data, so what does that really mean? What about self-learning algorithms? And what is that ultimately going to look like?”