Give a business owner an empty catalog and watch nothing happen

I'm generally a best-practices person. I think there's wisdom in the consensus positions that the data governance community has built up over decades, and I try not to be contrarian for its own sake.

But some recommendations read like they were written for an organization that doesn't exist. Nowhere is this more true than in data cataloging guidance.

Two things in particular.

1. The stewardship model assumes business owners know their data

Seiner and others have made the case (and I agree!) that Data Steward shouldn't be a job title. Stewardship is a responsibility that belongs to the business. The business owns the data; the business should own the metadata.

That's the right principle. Here's the problem with how it plays out in practice:

The people you identify as business owners and stewards are almost always drawn from the pool of people who work with the software sitting on top of your databases. Your application SMEs. And those people, more often than not, have never seen the underlying data. They know what the field is called in the UI. They have no idea whether "Name" in that single pane maps to ProperNameCD in the BusEntities table or Name_ID in the Business table.

The result is that your IT and data offices — the ones the model explicitly doesn't want in a stewardship role — end up having to occupy one anyway. Not because they're ignoring best practices, but because there's no one else who can actually answer the question.

The more realistic framing isn't "business owns stewardship." It's steward-partnership: a deliberate, documented handoff model where the data office fills the gaps the business can't and works to shrink those gaps over time.

2. A blank catalog is not a resource

Now for the part nobody wants to say out loud: when you acquire a data cataloging solution, your data office needs to jumpstart it before handing it to anyone.

I know that sounds like it violates everything I just said about stewardship. It does, a little. I'm fine with that.

An empty catalog handed to a business owner with a training session attached is not an asset. It's a mandate. And mandates without momentum stay empty.

What actually works is getting the metadata population percentage above a meaningful threshold (I think 20% is usually satisfactory) before anyone outside the data office is expected to engage with it. That means a few dedicated metadata jam sessions: the right people in the room, a focused scope, and someone accountable for keeping the catalog current as the dataset volume grows. In my experience, that's at least 10% of a staff member's time on an ongoing basis.

The psychological shift this creates is not trivial. When a business owner opens a catalog and finds something there, they explore it. They start to see the value. When they open it and find nothing, they close it and don't come back.

The gap between theory and implementation

The best practices aren't wrong, exactly. They're just written for a world where business owners already understand their data well enough to describe it, and where a freshly-deployed tool generates its own adoption. Neither of those things is true in most organizations I've seen.

The governance model you implement has to account for the org you have, not the org the framework assumes. That usually means the data office carrying more of the early load than the textbook recommends, and being deliberate about how and when you transfer that load to the business.

That's not a workaround; that's implementation.