MDM: Highly recommended, still misunderstood

Master data management is a hot topic.  And, this is at times surprising to me because the noise of big data is deafening.  Big data is certainly sexier.  MDM is like mom nagging to clean up the room - necessary, but a total buzz kill.

Here is some of the anecdotal evidence that is raising my eyebrows:

  • Our Forrester Wave for MDM was at the top of most read reports during Q1.  
  • MDM inquiries from clients keep me very busy.  
  • Vendors see MDM as a key growth area in their portfolios.
  • Consultancies are consistently pointing to client gaps in data governance and data architectures that point toward a master data problem.  

But, here's the rub.  MDM is thrown around like a silver bullet to master data problems.  It is a single box that looks so innocent in a reference architecture with names like product, customer, location, etc.  Data architecture recommendations add it as a bullet, something to check box in a mature practice.

Even as companies have implemented MDM packages or home grown solutions, the success stories are limited.  MDM runs in the background, forgotten about until the business has changes to the way they look at their data or execute a business process, or worse, when the maintenance bill comes due.  It has fallen into the trap of deploy and forget. Where is impacts the business is rarely if ever understood.  It seemed like a good idea at the time....  

So, why the renewed drum beat?

It's because of big data.  Not the volume of data.  It is the volume of data sources, the variety of data, and the speed at which data comes in and changes.  Managing a customer isn't simply to maintain the golden record in a CRM application or data warehouse.  Data silos are becoming networked organisms across distributed filesystems. databases and in-memory grids.  How can you orchestrate a consistent trusted view of master data without a master data management solution?  That team of data remediation specialists certainly can't keep pace with the task.

What concerns me is that MDM is still misunderstood.  Even as companies take the recommendations to implement it, they don't really know what they have. They want to load data into the hub, standardize the view, then push the data.  Huh, sounds like ETL.  Thus, how do you justify the value?  

I'll be honest, if you don't have to implement MDM - don't!  You need to weigh your options carefully.  "Why" is not too hard to guess.  But, it is important to repeat in case you are ready to just take the plunge and buy a tool and compliment of tools.

  • MDM is expensive.  
  • It is resource intense.  
  • It is implemented over months if not years.  
  • Most organizations don't have the right skills.
  • It has taken down data professionals before, and it will do so again because ROI is ambigous.  

MDM is not a check box in your data strategy.  Master data management is a data strategy, and not just for traditional environments. It matters for big data too in order to apply context to that data lake, making data useful in analytics and real-time engagment across a variety of engagement scenarios.

The value of MDM is context at automation and scale.  It provides the foundation for your enterprise data model.  The model that ate your business is tamed.  MDM can more easily brake down the the semantic and logical models, contextually syndicate proper views, maintain links and consistency between the views and how the data is used, and become another system of record to more easily explore data relationships and attributes of important data domains.  

If someone recommends MDM, be wary.  Be very wary.  Don't buy the tool before you know:

  • The domain model
  • The scope of consistency across systems and business views
  • The scale of data sources to harmonize
  • The quality of the data coming into the MDM tool
  • The standards data has to comply with
  • The number of buisness values it will support
  • The topline business impact to achieve

MDM is for quality, consistency, relationships, scale and automation of master data definitions and views, all for the sake of personalizing data for unique and varied business scenarios.    That is a tall order.  Are you ready?  


Can't agree more

Every statement rings a bell. It's a big investment in time even from the top leaders in the organization and in committed / skilled resources that bring both technical and functional knowledge to the table for the domain and industry in question.

Unless there is a separate team tackling this initiative one can forget about the ROI.

It is just not another initiative and a check box in the data strategy.

Master of none

A great post for MDM, but wanted to introduce a slightly different consolidation of acronyms for consideration relative to rational data management.

If the system architecture is highly structured with intent to capture highly specific, high quality data, we can then fairly easily provide a very strong ROI in many cases. I don't call our system MDM because it isn't like the systems out there that use the term, and our focus is business outcomes. However, the driver is often a combination of regulatory, governance, and related ROI scenarios--crisis prevention, accelerated discovery (as in science, disease, relationships, etc.).

Some have considered our system MDM, some BI, some productivity, others AI -- in fact it's in part all of the above. Indeed my goal as architect was to eliminate what customers don't need--yet often pay for, and focus on what's critical.

If you haven't taken a look recently, might find it interesting.

The rest I am in total agreement with -- don't underestimate enterprise-wide data management, either on the cost or the opportunity side.

Value is quantifiable in domains such as Materials/ Items MDM

Michele, you raise a fine point here. I have to agree with you that in many forms, MDM comes across merely as a must-do for a mature company, but value from it seems hard to find due to the deploy and forget trap.

At Verdantis however, we have never heard this grouse from our customers who engage in Item/Material master data management (MMDM) exercises. This in one of those few domains, where better management of master data is directly quantified into hard-dollar savings, be it due to 10-15% reduction in inventory (typically), identification of cheaper form-fit-function duplicates based on identical technical attributes, or even simply larger on-contract buying volumes.

In fact one of our customers - a G1000 manufacturing company - could directly point to USD 12 million savings in the first year after deployment of MMDM best practices, owing to above reasons.

With regards to MDM being expensive, resource intense, etc, there is immense truth in it when generic, one-size-fits-all solutions are forced on data from all domains like customer, vendor, material, etc. But, an MDM exercise quickly turns into a value multiplying initiative, completed in a matter of months, when end-user companies work with domain experts, rather than just large marquee IT solution providers.