Capitalizing on the Data Economy: Why MDM and Data Governance Have Never Been More Critical

Article Featured Image

After a few years of covering data governance (DG) and master data management (MDM) as an analyst, I began to receive questions from more than a few thoughtful architects and CIOs along the lines of “How did we get into this mess with our data? Were we all somehow negligent in regard to planning for, and taking a proactive approach, to data architecture and quality?”

These insightful queries caused me to step back and really think about the causes of the apparently ubiquitous need for data governance and management, and whether and why they will be required over the longer term. Many in our field simply take these concepts for granted, but in dissecting these issues and my thousands of interactions surrounding them, I discovered that their import is much greater than most realize.

From post-industrial to present day: Different technology, same problem

In the post-industrial revolution business world (and I do think it helps a bit to go back this far), executives were almost uniformly incentivized to optimize one or very few business processes. Larger enterprises began to develop multiples of these, and as they expanded, they acquired or promoted more executives. They paid them to optimize their processes in isolation, without regard to the methods the other executives were responsible for. In one form or another, all of this generated what we would now refer to as data silos in support of these varied operations. Any operational issues caused by data were either resolved, worked around, or ignored within those silos.

Fast-forward to the second half of the twentieth century, and the optimization of those processes began to mean IT automation. This was again done in isolation along business process boundaries, as business applications supporting single business processes came into being. Even implemented in this siloed manner, the operational efficiencies gained during this era cannot be overstated. If a data anomaly was serious enough to cause operational issues, they were approached in the same ways as they always had been—within each application silo. Here again, there really was no economic incentive to do otherwise.

With the onset of real ERP suites in the 1990s, there came the promise of automating most of the operational processes an enterprise required. However, the data required to perform these functions was still largely managed in silos within the application suites. Moreover, there were still functions like customer relationship management (CRM) that seemed to require isolated applications, mostly due to specialized business process requirements.

The wrench was thrown into all this automated bliss by the advent of analytics platforms such as data warehouses and data marts in the 1990s. The according nascent attempts that were made to bring the data together from these various application systems’ and suites’ data stores (abetted by the large-scale adoption of relational databases and the SQL programming language for both operational and analytical systems) brought disparities in data quality and semantics across these systems into stark focus.

The new “data analysts” attempting to gain even rudimentary insights discovered that were there serious issues with the data quality within each application. They also found that the implied definitions of corresponding data entities and attributes across these silos were often sufficiently disparate to render their efforts at analytics inaccurate or outright useless.

As the value of analytics had yet to be acknowledged, especially in comparison to the long-documented operational efficiencies that the business applications had brought, these data analysts required ways to correct these issues downstream from the sources of the data. This need resulted in the formation of the ETL solutions market. These data integration tools enabled data to be cleansed to some degree before being deployed to analytics platforms, but they were used without any management or governance, resulting in yet another set of data silos.

Why is all of this important? Because as enterprises seek—or are forced—to engage in digital business transformations, the approaches described above cannot continue as standard IT and business practice. Unfortunately, this way of doing things has forward inertia going for it that most don’t want to admit. The poster child version of this scenario is easily found in several companies that I’ve consulted with over the years.

These firms had so many of these silos that the projected or actual increased fixed costs in creating more of them in support of new products and/or services absorbed any additional revenue that might be realized by them. As awful as this effect of the law of diminishing returns sounds, and as obvious as it would seem to be something that should be prevented at all costs, I believe that most midsized and large corporations and the operating units within them are approaching some version of this barrier at greater or lesser speed.

How MDM is key to a successful digital transformation journey

The reason that “data and analytics” has recently gained the spotlight more than ever before, is to find ways to leverage the corporation’s accumulated data assets across these silos in support of new processes and analytics. That’s easier said than done as anyone who’s seriously attempted it knows, but the evolution of solutions such as master data management (MDM) and data governance solutions has now made it feasible. That is with sustained IT and business commitment to attack and resolve these issues in a scalable fashion.

Platform-style MDM solutions in particular provide the data model flexibility to allow the business to develop a common data model, often for the first time in an enterprise. These models accurately describe both the current and desired future state of the business and provide a map to those silos that are most critical to new and improved business outcomes along the digital transformation journey. With the formation of a virtual data governance organization, the business assumes responsibility for the state and use of the organization’s data while IT enables the managed deployment of the newly trusted data provided by the MDM platform.

As a result, the risk of employing data of low quality in the service of more modern, agile business processes is significantly reduced or outright eliminated. More importantly, the possibilities for digital transformation are limited only by the collective imagination of the enterprise itself.

KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues