What structured data actually does for a model
Structured data is markup added to a page that a visitor never sees and a machine reads directly — a way of stating who or what a page is about in a form built for parsing rather than persuading. It does not make a page more convincing to a person, and by Google's own account it does not directly improve where a page ranks either. What it removes is the guesswork a machine would otherwise have to do to work out what an ordinary paragraph is actually claiming — and guesswork is exactly what an assistant, working from a page it retrieved rather than a person reading it slowly, has the least room to do.
What is structured data, in plain terms?
Ordinary web copy is written for a person to read in context, with meaning carried partly by layout, tone, and the sentence that came before it. Structured data restates the same facts — a name, a role, an organisation, a date — as labelled fields a machine can read without needing that context, the way a form field is easier for software to process than a sentence describing the same information in prose. It sits alongside the visible page rather than replacing it: a visitor sees the ordinary copy, and a machine reading the page's underlying code finds the same facts again, tagged.
The vocabulary behind most structured data is not new, and it did not arrive with the current wave of AI products. Schema.org, the shared set of tags most sites use for it, was launched in June 2011 by Google, Microsoft and Yahoo, joined later that year by Yandex — four competing search engines agreeing on one common way for a page to state its own facts, so a site only had to mark itself up once. It has been maintained as shared infrastructure ever since, which is why the same markup a search engine has read for over a decade is also legible to language models built long after it was written.
What does it actually change for a search engine?
Google's developer documentation is direct about the limit: adding structured data to a page does not guarantee it will rank higher, and it does not guarantee that any specific enhanced result will appear even when the markup is written correctly. What the same documentation says structured data does do is help the search engine, in its own words, understand the content of a page and gather information about the people, organisations or other entities described on it — and it makes a page eligible for the richer result formats built on that understanding. Eligibility and understanding are the honest claims; a ranking boost was never one of them.
Does it work the same way for an assistant?
Structured data does not work identically for an assistant, and the nuance is worth stating plainly rather than smoothing over: Google's own guidance for its generative AI search features states that structured data specifically is not required for them, and that there is no dedicated schema.org markup built just for that purpose. What the same guidance says those features do need is a page whose claims are stated clearly enough to be retrieved and quoted with confidence — which is the same underlying problem structured data was always built to solve, even where the exact mechanism reading it differs. A declared identity and a plainly stated fact make a page easier for both a crawler and a model to draw from; formal markup is one way of stating that plainly, not the only one.
What should a non-technical buyer take from this?
The point of structured data was never to trick a machine into ranking a page higher — that claim was never true, on Google's own account, and repeating it here would be exactly the kind of overstatement this practice depends on not making. Its real value is narrower and more durable: it removes ambiguity about who or what a page is describing, for any reader that has to work that out automatically and at speed — a crawler, a model reaching for something to quote, or eventually a person skimming the same page in a hurry. A page that states its facts plainly, in a form built for parsing, gives all three the same unambiguous account to work from.