The question doing the rounds right now is whether schema is dead, or worse, a myth. The honest answer is more useful than either side of that argument. Schema never did the one thing people most often credit it with, and it still does several things its critics now overlook. Both halves matter, and holding only one of them is how the noise starts.
The myth worth killing: schema is not a ranking factor
Schema markup has never been a direct ranking factor, and Google has said so consistently for years. Adding structured data does not, on its own, move you up the results. If anyone has ever told you “add schema, rank higher”, that was wrong, and it is the claim most worth retiring. We hold ourselves to that too: schema is legibility, not leverage.
What schema does is make your content legible to machines. It states, in a form a parser cannot misread, what a thing is: this is a product, this is a person, this is an organisation, here is the event and its date. That is extraction and disambiguation. It is not ranking, and it never was.
What the “schema is dead” camp gets right
Some of the visible payoffs people associated with schema have genuinely gone. HowTo rich results were dropped. FAQ rich results were narrowed to authoritative domains around 2023, then removed from Google Search entirely on 7 May 2026. If your case for schema rested on those search enhancements, that case is weaker now, and pretending otherwise is the overclaim in reverse. We covered that specific change in detail in FAQ schema did not die, the rich result did.
What it gets wrong
“Schema is dead, full stop” is the same error as “schema ranks you”, committed from the other side. It takes one withdrawn feature and generalises it into a verdict on everything. Several structured-data types are very much alive and still earn rich results: Product, Review, Event, Video and Breadcrumb. And the entity layer, Organization, Person and sameAs, does work that has become more important, not less.
That entity work is the part the dismissers miss. It tells search engines and AI systems which entity you actually are, distinct from everyone else with a similar name, and links you to the references they already trust. It is how you become eligible for the Knowledge Graph. None of that is a ranking lever. All of it is identity and legibility, and identity is exactly what AI systems lean on when they decide whether they can name you with confidence.
What schema actually does now
- Disambiguates your entity. Organization, Person and sameAs say who you are and which references corroborate you, so a machine does not confuse you with someone else.
- Earns the rich results that still exist. Product, Review, Event, Video and Breadcrumb continue to produce enhanced search appearances.
- Makes your content cleanly extractable. Well-formed structured data helps the systems that ground AI answers read and quote you accurately.
- Connects you to corroboration. Linking to Wikidata, Companies House and official profiles via sameAs is how machines verify the claims you make about yourself.
- It does not rank you. That was never its job, and expecting it to is the oldest schema mistake there is.
Why the schema story feels so confusing
There is a tidy way to see it: the machine-readability frontier moved, it did not disappear. Google Search will tell you, plainly, that you do not need special files or markup to be found. In the same period, Google Chrome shipped an audit that checks structured signals and an llms.txt file for whether AI agents can use your site. Discovery versus functionality: being found is one job, being understood and used is another. Schema sits on the functionality side. Not how you get found, part of how you get understood. We unpack that split on the agentic readiness page.
So what should you actually do?
Keep your entity and your still-live schema. Do not rip out FAQPage markup in a panic, it is not penalised, it simply no longer shows a rich result. Stop expecting any of it to lift rankings. And judge each type by what it still earns. For the build detail, our schema architecture for the AI era guide covers how to do this well, from valid syntax to the semantic credibility that actually matters now.
The calm position, in one line: schema is not a ranking factor and never was, some rich results have gone, and the identity and extraction work is more valuable than ever. Mostly alive, with exceptions, which is the honest shape of almost everything in this field.
