Metadata as Common Language: An Onboarding Guide to Breaking Silos
Clean Data Architecture’s take on how “Fundamentals of Metadata Management” could become the onboarding guide we’ve all been missing.
Clean Data Architecture, with its most active writers Najate Bouad Engineer, Pierre-Yves Bonnefoy Architect and Gaëlle Seret Product, is about uniting the vision of different expertise and proposing a new cross-disciplinary take. In this article, Gaëlle proposes a complementary approach to reading Ole’s book “Fundamentals of Metadata Management”, without spoilers 😀. We hope it will spark even more ideas in the reader. Have fun!A long time ago, I was writing articles for my product consulting company. At that point, I was often joining new teams as Data Product Manager, and I kept encountering the same frustration: the onboarding was painfully incomplete.
I would get the usual information, the organisational chart, the stack, system access, tickets, maybe a bit of documentation, but I was missing everything that truly mattered to quickly build a coherent vision.
So I started writing down all the things I wished someone had told me:
The company meta-model.
The architecture overview.
The existing data domains.
Who was doing what, and why.
I turned these notes into a pragmatic article called “A Data PM Onboarding Checklist” and showed it to my manager at the time. He challenged me:
“Why do you need to understand the company meta-model?”
“Why do you need to see the architecture?”
“Why do you need to meet the architects?”
And honestly, I couldn’t demonstrate it. I just felt I needed those things, without being able to explain why.
This article resurfaced in my mind when I read Ole Olesen-Bagneux’s book “Fundamentals of Metadata Management”.
Here, finally, was the demonstration I had been missing all along.
Onboardings contain a lot of words but no meaning
When you join a new team, you are thrown into a web of systems, roles, vocabularies, and implicit assumptions. You spend weeks connecting dots that others take for granted. You’re not just learning where data lives: you’re learning what things mean. And that meaning is rarely documented, let alone shared across teams.
In most companies, every guild has its dialect:
Engineers talk about schemas and ingestion.
Product managers talk about metrics and journeys.
Architects talk about landscapes and domains.
Business talks about KPIs and ROI.
Everyone speaks, but no one really understands one another.
This is why the Data Product Manager onboarding feels so broken. Being transversal and working with a lot of different stakeholders, you’re the one completely reconstructing the shared language.
Metadata as the missing language
Ole Olesen-Bagneux’s approach is not solely for data engineers, it’s a guide for collective understanding.
He defines metadata not as a technical layer, but as context that travels with data. It connects meaning across systems, teams, and repositories. In his “Meta Grid” concept, he proposes linking all types of metadata: IT, data, information, and knowledge, so that organisations can build a shared map of meaning.
That map is what I was intuitively craving during those early onboardings. I didn’t want a better ticketing system: I wanted to see how things related to one another. I wanted to understand why something existed, who owned it, and how it fit into the company’s larger logic. In short, I wanted metadata.
From alignment to enablement
In large organisations, alignment is usually treated as a coordination problem. We argue that it is a language problem. As a Data Product Manager, this is the most important thing to tackle when you want to construct a data domain from scratch.
When I worked as a consultant in a luxury company, we had huge ambitions for the customer domain: helping brands to re-discover the face of their customers. We had no common tools, no central control tower and no massive governance board. Just an old centralised legacy and the vision for a transformation program.
What made it work was not hierarchy, it was shared understanding of business processes and pains across units and brands. Everyone, from data engineers to business leaders, started using the same words to describe the same things: domain, owner, data product, and source.
That shift transformed interactions. People didn’t need hand-offs to “translate” their requests anymore, they could reason altogether. And create a singular data strategy for the domain.
This is where Ole’s work meets Domain-Driven Design: both advocate for the creation of a ubiquitous language: a shared vocabulary built by the people who use it. Metadata is the operational form of that language: structured, discoverable, and reusable.
Toward a standard onboarding
A perfect onboarding curriculum built from Fundamentals of Metadata Management:
Understanding our company metadata
What are our main business contexts?
What are our data? What is our metadata?
How are ownership and stewardship practiced in our company?
Mapping meaning
Where is our metadata?
How do business, data, and IT repositories connect?
Speaking the same language
How to define common terms across teams?
How do we go from domain maps to shared glossaries?
Using metadata to enable collaboration
Do we embed context in tools, processes, and products?
How do we make discoverability part of our everyday work?
Designing your own Meta Grid
What do you already have?
What are the next steps you can take to build its next elements?
Such an onboarding would not only align newcomers, it would align everyone, continuously.
The real justification
So, to my former manager’s questions I can finally answer.
“Because understanding architecture is understanding meaning.”
“Because seeing the company meta-model is seeing how it thinks.”
“Because meeting architects is meeting the people who shape the grammar.”
We can’t build shared products without shared meaning. And we can’t build shared meaning without metadata.
The invisible structure of metadata is the foundation of every transformation effort.
Culture eats data for breakfast
Metadata is not bureaucracy. It’s culture.
It’s the quiet infrastructure that allows humans and machines to understand each other.
Ole Olesen-Bagneux has offered us a framework to break silos: to turn metadata from a governance afterthought into a common language of purpose.
So maybe the next time we onboard a new teammate, we can give them something better than a SharePoint link: we can give them meaning.
