What if Schema.org is just… Labels?
3rd November, 2024
We’ve all been sold the idea that adding Schema.org markup to your website will help Google to “understand” your content. Structured data has become SEO gospel, and I’ve been one of its loudest advocates. I’ve encouraged brands to invest in schema markup, to label their content, and to define their entities in order to unlock Google’s rich results and to improve their visibility. But here’s a thought: what if Schema.org is really just a glorified system for labeling things? What if it’s not some miracle of machine interpretation, but simply a method for adding neat little tags to our content – tags that Google may or may not pay attention to?
Don’t panic. I’m not suggesting that we abandon or remove our structured data; but it’s definitely time that we interrogated its actual role. The truth is, that Google’s usage of schema feels limited, selective, inconsistent, and often frustratingly opaque (and the data and reporting they provide is terrible). Most of it seems to only power specific rich results — think FAQs, reviews, product listings. Beyond that? It’s hard to tell if Google even “reads” it, outside of the trickle of new features emerging around Google Merchant Center and product listings. That leaves me wondering: is the broader schema landscape just window dressing?
If it’s just labels, does it matter?
Let’s assume, for a moment, that Schema.org’s utility is limited to acting as a labelling system. No magic, no “understanding,” just a way to tag your content with specific information. Even if that’s the case, it doesn’t mean it’s pointless. Labelling things properly is still a powerful act, especially when it involves repetition and relationship-building. When Schema.org markup echoes and interlinks key details, it reinforces intent, connects concepts, and gives structure to the relationships between entities. It’s the difference between scattering isolated facts and defining a network of meaning.
Imagine Schema as a framework – a skeleton – that’s consistent, logical, and machine-readable. That might be enough to make a difference, especially as other systems come into play.
Relationships as reinforcement
Schema allows us not only to label individual things – an “author,” a “brand,” an “article” – but to connect them in meaningful ways. The author writes the article; the brand publishes it. These connections build a kind of narrative map that both reinforces the importance of each entity and gives machines context for how they relate to one another.
When we create these connections in the schema, we’re establishing relationships that act as semantic “anchors.” Machines don’t just see an article on its own; they see information about who wrote it, who published it, and the interwoven story of how each element fits within a larger framework. This structure isn’t about directly influencing rankings, but it’s a contribution to clarity and consistency that ensures machines interpret something correctly – because relationships are often easier to interpret than standalone facts.
Even if Google doesn’t always use these connections to “understand” us in the way we’ve been told, they’re there, giving our content a connected structure that LLMs, knowledge graphs, voice assistants, and other emergent tech can grab onto. The web is increasingly navigated by non-human entities, and these entities benefit from roadmaps that define not just what something is, but how it relates to everything else. Schema helps us to create that map.
General understanding vs explicit intent
It’s no secret that LLMs and AI models are getting better at extracting context and understanding from content. They can process vast amounts of text, infer meaning, and generate responses that feel like they “get” the topic. But that’s fundamentally different from schema’s role. Schema.org isn’t about broad context; it’s about the specific intent and emphasis defined by the author. It lets us say, “This is what this page is really about”, with no ambiguity. Schema adds an extra layer of precision, signalling to machines not only what’s present on (and/or which can be inferred from) a page, but what the author considers essential.
This explicitness helps machines pick up on the nuances and hierarchy of importance that pure content analysis often misses. Without schema, an AI might grasp what a page is about, but schema allows us to direct it to the precise elements, relationships, and priorities we intend to highlight.
Looking beyond SEO
We’re moving into an age where search and discovery are influenced by AI as much as they are by traditional search engines. Structured data is our way of putting a signpost on every page, every image, every fact, and every product, with labels that also describe how they interconnect. AI doesn’t “look” at a website in the same way that a person does; it sees data points in a web of relationships. Schema helps to make sure that what’s visible in that layer is consistent and, ideally, correct.
And when those relationships are repeated in the markup, the machine becomes even more certain.
Imagine the power in ensuring that the machines parsing your page can see not only which entities are important, but how they fit together. It doesn’t take a visionary to see that structured data will be a backbone for tomorrow’s digital ecosystem. When voice assistants answer questions, when LLMs like ChatGPT pull information, it’s this structured data that lets them make connections without human intervention.
Practical strategies
Of course, there’s no point in blindly implementing schema for schema’s sake. Be deliberate. Prioritise “actionable“ schema types, like FAQs and Product markup, for the SEO wins we know they bring. But don’t shy away from building a connected graph of broader, “descriptive” schema just because Google’s not showing an immediate return. These “descriptive” types and relationships might end up being the lifeline between your content and the AI models of the future.
Think of it as building a data-first foundation, one that’s adaptable, connected, and machine-readable, not just today but in the landscape that’s emerging. Don’t add markup without thinking about the purpose of each field and the relationships it defines. And don’t forget that this isn’t a set-and-forget deal. The machines reading our pages are changing, and so are the requirements of the systems that interpret schema. Iterate. Refine. This is an ongoing investment in your content’s long-term legibility.
The long game
Yes, maybe Schema.org is “just” labels and connections. Maybe we don’t get any magic powers from it right now. Maybe we were over-sold by Google, and maybe that goes some way to explaining their mixed messages about the value and impact of structured data.
All of that aside, it’s still worth implementing, maintaining, and paying attention to. Schema.org enables you to make a website that’s legible not just to Google, but to the broadening universe of AI and machine-learning systems that are becoming our primary digital gatekeepers. Schema is less about instant returns, and more about future-proofing. It’s an investment in clarity, in digital durability, and in keeping pace with the world our content is moving into.
Whether Google “gets it” or not, machines need signposts and maps of relationships, and we’re putting them up. The future of search, of AI, of discovery, isn’t waiting for us to be ready – it’s here, and Schema.org might just be the baseline that keeps your content in the conversation. So let’s keep labelling things and defining how they connect. One day, the machines will thank us.
Man!!! You got me with that.…yeah, title! Loved it, especially this line “Schema is less about instant returns, and more about future-proofing”. Thanks for sharing your perspective. 🙌
I’m okay with building it into it, or even better, have a frontend that has it onboard. But proving the business value of anything more that ‘prettier results in Google’ is a real pain in the *ss. People don’t like to stuff for fun or good these days. It’s a shame.
Thank you for your thoughtful input.