What centralization needs to do is bring a diverse and democratized set of ideas and thoughts forward, so that the entire university benefits from that voice. Central IT is enabling that voice to meet the objectives of the university mission. Higher ed IT leaders should think globally and allow areas to act locally. If what we really want is command and control, things will break down, because it’s very difficult to scale a central organization to meet the complete needs of the institution. And that’s where federated governance comes in.
When Bright People Work With Structure: What ‘Pockets of Excellence’ Really Means
In universities, you’re always going to find very bright people who are very capable of doing things with data. And so, without much structure or guidance, you end up with what I would call “pockets of excellence.”
Someone in the registrar’s office builds fantastic reports. A department chair has a homegrown data mart. An advising center figures out a way to identify students at risk academically faster than anyone else on campus.
That autonomy with structure is the heart of a federated data governance strategy. Central IT’s role is to create the structures, inclusive of definitions, tools and guardrails, to enable these pockets to operate with more impact, not less freedom.
LEARN MORE: See how Notre Dame University employs a flexible data management strategy.
The Difference Between a Data Owner and a Data Steward
A lot of governance efforts fail in a very specific way: Stewards are “volunteered” in name only.
Somebody gets their name put on a slide as “data owner” for a domain they don’t really feel accountable for, in an area where they haven’t been given the time or training to do the job. That’s not stewardship that sets up that individual or the overall governance for success.
When I think about data stewards, I’m looking for a blend of people — people who have a passion to work with data and those who understand the calculus of good data structures to create better outcomes. You don’t need all of one group. You need people who understand the impact of improper use and the value of proper use within their context, their domain and their understanding.
Somebody who is very knowledgeable on data but doesn’t get the business needs or the bigger picture of privacy and security and integration might not be as valuable as somebody who gets the bigger picture and can be taught the data concepts. We can teach the data concepts. It’s harder to teach institutional context, judgment and trust.
So, stewards need to be named on purpose and trained on purpose. That’s the difference between a governance model that shows up in quarterly minutes and one that shows up in everyday decisions.
Early Warning Signs of Data Governance Breakdown in Higher Education
People sometimes frame this as an “AI problem,” but it’s really a centralized data problem. When AI starts consuming institutional data across distributed systems, a model trained on bad data from one department affects outcomes for everyone. But that’s just shining a brighter light on issues that were already there.
If you have governance, it becomes self-governing. The real challenge is the new data coming in — how it’s requested, captured and defined. The early warning signs are pretty consistent: Someone goes rogue on how they ask for or create new sets of data that already exist. People stop using the gold copy and start rebuilding local versions “their way.” The same person shows up twice because two systems asked for “name” differently. Those are not just technical problems.
Those are governance problems. And they’re exactly what federated stewardship is meant to catch early, before it becomes a high-profile data quality incident or an AI failure.
What Makes Data Governance Actually Self-Sustaining
I don’t spend a lot of time arguing about where data has to live. I just believe that there has to be an owner. Otherwise, nobody owns it.
I’ve seen a lot of different structural models work, but in every case where governance is successful, it’s because it’s a university initiative that is supported at the highest levels. Once governance is established and vetted by the group with that authority, it can become self-governing if it’s simple enough to use.
That includes clear guidance on how to ingest data, how to operate on data and how to expose data in ways that respect privacy and security. When you do that, it works.
Partners such as CDW can help make that case and translate it into practice: listening, bringing cross-sector perspective and helping institutions think through what to hold centrally, what to distribute and how to stand up governance that operates every day.
In that world, the CIO’s job is not to own all of the data. The CIO’s job is to make sure the right people own the right data, within a structure that the whole institution can actually use every day, not just every quarter.
