The Godfather of Data Governance: Bob Seiner on Data vs AI Governance, and The Data Catalyst Cubed
TL;DR
Bob Seiner says most companies already govern data — they just do it informally — his core idea from “non-invasive data governance” is to formalize accountability people already have, instead of dumping new responsibilities on them from the top down.
He rejects both command-and-control governance and “if you build it, they will come” programs — Seiner argues those two common approaches fail, while recognizing existing stewards and embedding governance into day-to-day work is what actually gets adopted.
His new framework, “The Data Catalyst Cubed,” is an equation: data governance × change management × data fluency — the point is that good data alone isn’t enough; organizations also need behavior change and people who are confident and competent using data, especially in AI-heavy environments.
Seiner draws a hard line between data governance and AI governance — data governance is about authority over the definition, production, and use of data, while AI governance deals with the behavior and outputs of models, and he says companies need plans for both.
The AI era makes old governance problems feel newly dangerous — Joe Reis and Seiner trade examples like Copilot generating analytics from “whatever steaming garbage pile” lives in SharePoint, or employees uploading sensitive finance and healthcare data into LLM tools without clear rules.
Seiner’s origin story goes back to Blue Cross Blue Shield in the mid-1980s and Larry English’s article “Accountability to the Rescue” — that article, plus early work in metadata and stewardship, led him to become “Mr. Metadata,” launch TDAN in 1997, and eventually trademark “non-invasive data governance.”
The Breakdown
The joke-filled intro and the case for “non-invasive” governance
Bob Seiner opens like someone who’s been around long enough to laugh at himself, joking about finally finding his one fan before explaining he’s been doing data governance since before it was even called that. He traces his thinking back to stewardship and Larry English’s “Accountability to the Rescue,” then lays out his now-signature point: companies are already governing data, whether they admit it or not.
Why traditional governance programs flop
Seiner says there are really three approaches in the wild: command-and-control, the “Field of Dreams” version where you build a program and hope people show up, and his non-invasive approach. His critique is simple and sharp: don’t assign people extra work and call it governance — recognize the data responsibilities they already carry and formalize them.
“Data enablement” gets a hard no
When the conversation turns to rebranding governance as “data enablement,” Seiner flatly calls BS. To him, “enablement” is too passive, while governance should be a catalyst — the spark that accelerates BI, analytics, and AI instead of a cutesy label meant to make people less uncomfortable.
The Data Catalyst Cubed equation
This is the centerpiece of the episode: Seiner’s new book, The Data Catalyst Cubed. His formula is data governance × change management × data fluency, and he stresses the multiplication matters — if you don’t change behavior and help people become confident and competent with data, governance alone won’t carry the load.
AI raises the blast radius
Joe Reis brings the modern anxiety: finance teams wondering what they can safely upload, healthcare orgs dealing with HIPAA risk, and employees using Copilot like a mystery BI tool over random SharePoint files. Seiner’s key point is unsettling but useful — even if the data is pristine, documented, and high quality, people may still not know how to use it correctly, which is why behavior and fluency matter as much as control.
Data governance is not AI governance
Seiner says he recently wrote a deliberately clickbaity article claiming data governance and AI governance “are not related,” then explains what he means. Data governance covers authority over data itself — how it’s defined, produced, and used — while AI governance is about the behavior, decisions, and outputs of models; separate disciplines, both essential.
The Bob Seiner origin story: Blue Cross, metadata, and 1984-era stewardship
His entry into the field came through Blue Cross Blue Shield in Pittsburgh, where a graduate school classmate recruited him into a “data consultant” role. That sent him deep into metadata management, repositories, and modeling, and eventually into stewardship work after a first chief data officer started asking the classic question: “Who owns the data?”
Building an industry voice with TDAN and a differentiating idea
Seiner explains that pushback to “non-invasive” was real — some people told him governance had to be invasive to work. But he stuck with the phrase because it captured something memorable and practical, later even trademarking it; the same instinct drove him to launch TDAN in 1997, back when publishing online meant FrontPage, HTML files, and physically photocopying magazine articles to share ideas. By the end, his advice for anyone building a reputation is very Seiner: find the thing that genuinely sets you apart, because otherwise you’re just more noise on LinkedIn.