Note: This is the second of three articles in the series “Data Governance (Should be) the Next Big Thing. It is co-written by the excellent Monica Fulvio.
Common Pitfalls in Data Governance
Understand that data governance is first and foremost an organizational problem, not a technical problem. The options for solving governance with technology are limited, even in the age of LLMs. You can build procedures and systems to maintain a standard or a common understanding, but you can’t build a system to create an understanding out of disparate data–at least, not an understanding that will be adopted and followed.
Organizations struggle with data governance when there is a lack of strategic clarity, based probably on unrealistic expectations of effort, scope and cost of governance efforts. As we noted in the first part of this series, data governance isn’t a one-time cleanup effort, or a one-time modeling effort. It is day-in-day-out vigilance, enforcement and assessment. It takes people, systems, organization and leadership to make it work. The old “let’s put all the data in one place and figure it out later” playbook doesn’t work. So, in identifying pitfalls, naivete about the effort required is a big culprit. We’ll come back to that topic.
Another major issue is the misunderstanding of scope. Here there is good news and bad news. The good news is that there is very likely some element of data governance going on all over the organization. After all, the users of your SAP system probably don’t worry about data governance in terms of quality or format, because SAP does it for them. Salesforce users don’t spend a lot of time worrying about the data model of customer data or whether it’s compatible with data from other systems. Compatibility is someone else’s problem. The managers of your CMS probably don’t care much about whether the metadata and taxonomies they create are in line with upstream systems, because they produce data first and foremost for presentation to the public. But each of these users care about and understand the data they create and work with, even if they are likely doing so within their own sphere and silo.
And there’s the key: corporate governance strategy is concerned with accuracy, consistency and security at the point of origination, and access, compatibility and automation at the point where data crosses from one silo to another–that is, in the sharing and use of data for other than its original purpose. The scope of the effort has team, business unit and corporate aspects. Given the ubiquitous use of specialized data systems like SAP and Salesforce, there is always an element of translation or transformation when data from such systems needs to be shared. That’s the bad news: There isn’t a one-size-fits-all model or approach to governance. Good governance recognizes and manages data based on its intended use. Further, that needs to happen at every level within the organization. The scope, therefore, of data governance spans data creation, acquisition, sharing and reuse.
Given these common issues, let’s take a look at some specific points of failure for data governance efforts, whether they never get off the ground or lose steam over time. These are based largely on observations over decades of personal experience, but continue to be issues today.
“Just Catalog It”
While data cataloging is an essential effort in data governance, it isn’t an end goal by itself. And sometimes, data governance starts and stops at discovery. Catalogs allow organizations to understand what data they have and what it means. They are purely cost centers if not supporting a larger goal. Used with a broader goal in mind, a data catalog can unlock and support groundbreaking research, help people find vital resources that might be hard to find otherwise and identify gaps in data coverage. If an organization stops at “just catalog it,” the catalog is more likely to go unused and gather dust or eat up resources until it’s abandoned.
Marathon Not (Just) A Sprint
Another place where we’ve seen governance efforts fail is curtailing support, and then seeing a lack of success from an undersupported program. Curtailed support can take a variety of forms–not only lack of staffing or budget for the governance team directly, but often more perniciously, a lack of engagement, time and support across stakeholder teams. Successful governance requires engaged stakeholders and stewards, which in turn demands stakeholder teams be generous with their time in support of the effort.
While it’s wise to plan on interim short-term wins, especially as you build out your governance program, data governance is a vital piece of solving your data problems and advancing your data strategy; just as you’re unlikely to achieve your data strategy in a year or two–and if you do, it’s time to set your sights higher–your governance program will need to be an ongoing, evolving investment.
Lonely Data Governance
Another common error is looking to data governance alone to solve your data problems. No matter what kind of data problem your organization is trying to address, it requires some mixture of different data management disciplines (e.g., data modeling, data architecture, reference and master data, data quality, security, metadata management, data operations, business intelligence) to engage, where data governance should be a central and coordinating hub function, but cannot replace these other functions.
How often have you found a governance team and a recommended standard well after embarking on a project where you’ve already created data in a different format? These issues come back to organizational failures: lack of communication, lack of enforcement of a standard, etc. Sometimes there’s a lot of pressure to “just get it done,” which is shorthand for “I don’t care if the data works for other teams or will slow us down in the long run and lead to compliance and other issues. I have a deadline!” Leadership should be strong enough to overcome such short-term corner cutting. More importantly, if the data governance effort and outcomes are baked into data creation and management pipelines, new projects can start up with having to think too much about compliance with an internal standard. And, your data governance team won’t feel like they’re doing make-work.
Change Is Hard and So Are Organizations
Ongoing engagement and support needs to include strong leadership support and also engagement and support among stakeholders working in the data trenches. It’s easy to focus on the (very different) communications and relationship management with one audience or the other, but both are essential to ongoing governance success.
One of the reasons that both grassroots and executive support are essential is that at the heart of every successful data governance program is a thoughtful approach to organizational change management: At heart, governance asks us how we can better manage our data together, do so more reliably, responsibly, ethically, and efficiently; in short, how we can improve our data maturity as an organization. And doing things better inherently requires doing them differently, where some of those changes may be uncomfortable and require focused effort. Understand and plan for organizational change management as a core part of your governance work.
Being Immature
So you did a data discovery effort, right? And surely as part of that you also did a data management maturity assessment (DMMA), yes? A long-term data governance effort requires a long-term commitment to improving and then maintaining a high level of data maturity.
Again, this isn’t about the systems you choose to integrate–data catalogs, master data management platforms, medallion data pipelines–it’s about recognizing that good data is good business. Many companies start out to improve their data with great zeal, then realize it’s a lot of effort and, never having built the governance structures and measurement tools to ensure a long-term commitment, stop governance altogether or let it atrophy. No executive, after all, is going to say “I don’t care about data quality.” They are instead going to point to their (probably) anemic governance effort as a demonstration of commitment, ignoring whether the business is getting any measurable benefit from better data hygiene. So, immature data (according to standard models of assessment) is a sure sign that existing governance efforts are just window dressing.
In Summary
Hopefully you’ll vault clear over all of these pitfalls, and be on your way to a much smoother (if never entirely easy) path to great data governance.
And if some of these pitfalls feel stingingly familiar, you’re in good company; we certainly learned some of these the hard way in our own experience, and know smart folks at good organizations who’ve stumbled along the way. Each pitfall can be a learning experience; seeing these go wrong can tell you interesting things about your organization and data management efforts. But even the smoothest road to great data governance is long and complex–why not take a few wise shortcuts where you can and learn from us?
In our third and final installment in this series, we’ll conclude with some recommendations on how to approach the issues, with an emphasis on understanding your needs.