Ontologies are all the rage
Has the practice of ontology come across your desk yet? It's come across mine a few times, and lately, with increasing frequency. So as I do with so many other things when I want to really dig in, I thought I'd write about it.
Ontology might be the most important practice that content designers and strategists aren't talking about enough (yet) — and in the age of AI, that's becoming a real problem. Let’s get into it.
What is ontology?
Ontology is the practice of defining what things are, what they're called, and how they relate to each other. It's the difference between a list of seemingly random words and a shared understanding of what those words actually mean.
Sound familiar? It should. Us content folks do this for a living, no matter what we call it — creating consistent labels and clear terminology; offering shared language that means the same thing to everyone who encounters it. Ontology is just the formal name for that work.
Ontology by any other name
If you've ever worked with Object-Oriented UX (OOUX) — Sophia Prater's methodology for defining the core objects in a product before you design a single screen — you've already been doing ontological thinking. You just might not have called it that. When you work with OOUX, you identify what the real things are, in a system, how they connect. That's ontology.
The advantage offered by ontologies in organizations using AI is almost immeasurable these days — when content drift is happening everywhere, faster, and all at once. But when your content strategy is grounded in the same object model as your user experience, everything becomes more coherent. The UI labels, the help docs, the chatbot responses — they all speak the same language, because they're drawing from the same map of meaning. The same ontology.
A common scenario
In our roles as content practitioners, we face these familiar challenges all the time:
One product
Two (or more) teams
Three (or more!) different names for the same feature
A user who calls it one thing
A support doc that calls it another
A UI label that calls it something else entirely
And…A chatbot that can't answer questions because it has no idea what anyone is talking about
THAT’s an ontology problem. And not only does it lead to seriously degraded user experiences, it fractures customer trust, creates friction, and quietly erodes the credibility of everything you build. That makes it expensive, in more ways than one.
Why we need ontology NOW
As we all know by now, AI tools are only as good as the content they're trained on or connected to. Feed them muddled, inconsistent language and you'll get muddled, inconsistent outputs. Every. Single. Time.
But if your content has a solid ontology underneath it — clear objects, consistent terms, defined relationships — AI can use it to create higher-quality, more consistent outputs, while answering questions (think AI chatbot) consistently and accurately. That's where content practitioners come in — we define the ontology, the language patterns, that AI runs on.
Where to start
A content ontology is the foundation that everything in a software platform sits on. Tone of voice, style guides, content models: none of it holds together without shared meaning underneath.
Like anything new, it can feel daunting to start. But there's some good news on this front: working with ontology is probably something you’ve been already doing for years, just under another name. So go ahead, label it — and create an ontology today.
Here's how:
Audit your organization’s terminology. Pick one key concept in your product or organisation and document every name it goes by. The results will be illuminating — and probably a little horrifying. Hang in there — it gets better.
Use OOUX as a foundation. If you've already mapped your product objects, you're closer than you think. If not — now’s a good time to start. Make sure your labels, docs, and support content all reflect the same ontology.
Bring your output into your AI conversations and inputs. Ask questions like: How do we make sure we’re using shared language in our content models? What content are we using to train our AI? Who owns that content? How do we keep it consistent? These are content strategy questions — own them.
Build a glossary and make it relational. In other words, not just "here's what this word means," but "here's how this word connects to these other words." Lean into work you've previously done with OOUX, or make OOUX your new best friend.
Remember — this isn't a brand new skill you need to learn. If you've ever sat in a room and fought for consistent terminology, pushed back on a label that didn't reflect how users actually think, or asked "what do we actually mean by that?" — you've been doing this work already. You know this. And you know how to do this.
Ontology reinforces our age-old argument — content is not decoration. It is structural. It shapes how people understand products, make decisions, and trust organisations. The words we choose, the names we give things, and the relationships we define between them are not nice to have — they are a foundational piece of a product’s architecture. AI didn't create this need. It just made ignoring it a lot more expensive.
So keep going. Keep doing what you do. It matters more than ever.
Comments