I'll give the short answer first, because it tends to surprise both sides of the debate: schema markup still matters, but not for the reason most people think, and not in the proportion many people sell. If you were hoping for a "yes, it's everything" or a "no, it's dead," sorry to disappoint — the honest truth lives in the middle, and it's more useful there.
What Schema Actually Does
Structured data is a way for you to state, explicitly and machine-readably, what each piece of your page means. "This is a price." "This is a rating." "This is a frequently asked question, and this is the answer." Instead of the system having to infer everything from raw text, you hand it labels.
In the era of traditional search engines, that became synonymous with rich snippets — those stars, prices, and FAQs that appeared right in the results. The value was visible and direct. That's how schema earned its "mandatory" reputation.
Why the Role Changed With LLMs
Here's the nuance that changes everything. Modern LLMs are extraordinarily good at understanding unstructured text. They don't need schema to grasp that "from $99/month" is a price, or that a question-and-answer block is a FAQ. They read the way a human reads — and make sense of it without needing labels.
That knocks out the strongest argument propping up schema: "machines don't understand text, so you need to structure it for them." Today's machines do understand. So part of schema's historical value has, indeed, shrunk.
So Why Does It Still Matter?
Because "understanding" and "interpreting unambiguously, with confidence" aren't the same thing. Schema is no longer the only way for a machine to understand you, but it reduces the chance of a misinterpretation — and in a world where AI decides what to say about you, reducing ambiguity has value.
There are three concrete reasons it stays relevant:
Extraction confidence. When the price, product name, or rating is explicitly marked, there's less room for the system to get it wrong. The LLM can get it right reading the text, but schema lowers the odds of confusion.
Search systems still use it. Many AI answers pass through a traditional search layer before reaching the model. That layer still uses structured data. In other words, schema influences what gets retrieved, even if the final model doesn't depend on it.
It's cheap and has no downside. Implementing correct schema is one-time, low-cost work with no risk. When something helps at the margin and costs almost nothing, the math works.
Where Schema Was Oversold
Let's be honest about the other side. Schema won't make an authority-less brand get cited. It doesn't compensate for weak content. It's not a visibility button. Anyone selling schema as the secret to AEO is selling an illusion — it's technical hygiene, not strategy.
And there's a common excess: cramming the page with irrelevant or misleading markup in the hope of "signaling" more. That doesn't help and can hurt. Good schema is honest, pertinent schema, not abundant schema.
The Practical Position
Treat schema for what it is: a low-cost technical fundamental that reduces ambiguity and helps the search layer, not a growth lever. Implement what makes sense for your page type, do it right, and move on to what actually moves the needle — authority, reference content, presence in sources the AI respects.
In other words: do schema because it's cheap and helps at the margin, not because someone told you it's the heart of AEO. It isn't. The heart of AEO is being an answer worth citing — and no structured data builds that for you.
If you want to know whether your fundamentals, schema included, are actually improving how AI sees and cites your brand, the way is to measure the outcome, not the implementation. That's what Genoma tracks: not whether you marked up the page, but whether AI is understanding and citing you better because of your effort as a whole.