Schema markup has been on the SEO to-do list for years. Most marketers know they probably should be doing something with it. A reasonable number have set something up. And a surprisingly large chunk of those have set it up incorrectly, then wondered why it’s not doing anything.
Here’s the thing, though. In 2026, schema isn’t the quiet technical nicety it used to be. It’s become one of the most direct levers you can pull to influence how Google, ChatGPT, Perplexity, and every other AI-powered search experience understands and presents your brand. Get it right, and you’re feeding the machine exactly what it needs. Get it wrong, and you might as well be shouting into a void.
So let’s get into what schema actually is, why it’s shifted from a nice-to-have into a genuine strategic priority, and what proper implementation actually looks like in practice.
What Is Schema Markup, and Why Does It Exist?
Schema markup is structured data. It’s a standardised vocabulary of code, most commonly written in JSON-LD format, that you add to your website to tell search engines and AI systems exactly what your content means, not just what it says.
Without schema, Google has to interpret your content through natural language processing. It reads your page, makes educated guesses about what things are, and pieces together a picture. Most of the time, it does a reasonable job. But it’s still guessing. Schema removes the guesswork entirely. Instead of hoping Google figures out that the number on your page is a phone number, schema explicitly tells it: this is a telephone number, this is a business address, this is a product review, this is an FAQ.
The vocabulary itself is maintained by Schema.org, which was co-founded by Google, Microsoft, Yahoo, and Yandex. As of 2024, over 45 million web domains use Schema.org markup across more than 450 billion objects. It is genuinely the universal language of structured web data.
So What’s Actually Changed in 2026?
The shift that’s made schema suddenly much more important isn’t a Google algorithm update. It’s AI search.
Google’s AI Overviews now appear on a significant proportion of search queries. ChatGPT processes billions of queries every month. Perplexity is growing fast. And all of these systems share one important characteristic: they don’t read your website the way a human does. They’re pulling structured, machine-readable signals to build their responses, and they heavily favour sources that make their job easy.
A controlled experiment published by Search Engine Land put this directly to the test. Three near-identical pages were created, one with a well-implemented schema, one with a poorly implemented schema, and one with no schema at all. The result was stark. The page with a well-implemented schema was the only one to appear in an AI Overview. The poorly implemented page ranked for keywords but didn’t appear in AI Overviews at all. The page with no schema wasn’t even indexed.
That’s not a marginal difference. That’s the gap between existing in AI-driven search and being completely invisible to it.
Google’s own Search Central documentation notes that Rotten Tomatoes saw a 25% higher click-through rate on pages enhanced with structured data compared to those without. That data predates AI Overviews entirely. The advantage has only grown since.
Schema as an AI Readability Signal
This is the framing shift that matters most for 2026. Schema isn’t primarily about rich snippets anymore, though those are still genuinely useful. It’s about AI readability.
When Google or ChatGPT is deciding which sources to cite in a generated response, they’re looking for content they can parse with confidence. A schema is essentially a translation layer between your content and the machine. It removes ambiguity, defines relationships between entities, and confirms that what’s visible on the page matches what’s in the code.
That last point is more important than it sounds. AI systems actively check for consistency between the schema and visible content. If your Article schema says the post was published on one date, but your page shows a different one, that’s a red flag. If your Product schema lists a price that doesn’t match your actual product page, that’s a problem. Mismatched schema doesn’t just fail to help; it can actively signal that a source isn’t trustworthy.
The Google Search Central general guidelines are explicit on this: structured data must be a true representation of the page content. That sounds obvious, but a huge number of implementations fall down here, particularly on sites using plugins or templates that auto-generate schema without checking it against what’s actually on the page.
The Schema Types That Actually Matter Right Now
Google deprecated a handful of schema types in January 2026, including PracticeProblem structured data. This caused a minor panic in some SEO circles, with people wondering whether structured data was being phased out entirely.
Google’s John Mueller addressed this directly on Reddit, and Search Engine Journal covered it well. Mueller confirmed that schema types “come and go” but that core types remain essential. His exact framing was that “markup types come and go, but a precious few you should hold on to.” Google isn’t killing schema. It’s refining which types deliver the most value.
The types that continue to matter most in 2026:
Organisation is arguably the most underrated. A clean, complete Organisation schema on your homepage with your address, phone number, email, sameAs links pointing to all your authoritative profiles, and a properly nested AggregateRating is the single most important entity signal you can send. It tells Google and every AI system exactly who you are, where you’re based, what you do, and how trustworthy you are as a source. Plenty of sites have some version of this, but most are incomplete or use incorrect types (more on that in a moment).
Article and BlogPosting schema on every piece of content, with a named, attributed author linked via sameAs to their LinkedIn or other authoritative profiles. This directly feeds E-E-A-T signals. AI systems weigh author expertise heavily when selecting sources to cite, particularly for anything touching on business, finance, or professional services. If your content is published anonymously or without structured author data, you’re leaving that signal on the table entirely.
Product schema for e-commerce is non-negotiable. AI search tools are increasingly being used for purchase research and product recommendations. Research cited across multiple industry sources shows that 65% of pages cited by Google AI Mode and 71% cited by ChatGPT include structured data. For product pages specifically, every GTIN, MPN, brand, availability status, price, and currency field needs to be populated. An incomplete product schema is almost as bad as none at all for AI visibility.
FAQPage schema is still worth implementing correctly, even though Google pulled the FAQ rich result feature for most sites in 2023. The reason is that the FAQ and HowTo schema still helps AI systems extract direct answers from your content. The structured Q&A format is one of the cleanest signals you can give an AI crawler.
LocalBusiness schema, combined properly with Organisation, is essential for any agency or service business with a physical location or defined service area. This is what powers AI-generated local recommendations.
The Difference Between Having Schema and Having It Right
This is the bit that most guides skip over, but it’s where the real opportunity sits.
A significant number of websites technically have schema markup. Many of those implementations are broken, incomplete, or conflicting. Common issues include:
Combining Person and Organisation types on a single schema node. This creates ambiguity about what the entity actually is. Google wants a clean, unambiguous Organisation entity for a business. A muddled type declaration undermines the whole point.
Having a schema that doesn’t match the page content. If your WebPage schema references a dateModified from six months ago but the page was updated last week, that’s a mismatch. Keep it current.
Letting third-party plugins inject their own schema without checking what they’re actually outputting. Review platforms like TrustIndex, for example, often inject their own Product schema to carry an AggregateRating. That’s technically functional but semantically wrong for a service business. The AggregateRating should sit on the Organisation node, not a Product type.
Missing required properties. Google’s Rich Results Test will tell you whether your schema is eligible for rich results, but it won’t always flag every missing recommended property. Running your schema through both the Rich Results Test and the Schema Markup Validator gives you a more complete picture.
Schema and Your E-E-A-T Signals
One thread that connects schema to the broader SEO landscape right now is E-E-A-T. Google’s framework of Experience, Expertise, Authoritativeness, and Trustworthiness underpins how it evaluates content quality, and schema is one of the clearest ways to make those signals machine-readable.
Person schema on author profiles, with sameAs links to LinkedIn and other professional profiles, turns a name in a byline into a verifiable entity. Organisation schema with proper sameAs links to Companies House, Google Business Profile, and industry directories reinforces that Repeat Digital, or whoever, is a real, established business with a verifiable footprint.
This matters more than ever because we’ve written before about how GEO and AI search works. AI systems don’t just look at your content in isolation. They cross-reference it against everything else they know about the entities involved. A named author with a rich, consistent online presence is far more likely to get cited than an anonymous post, regardless of content quality. Schema is how you make that presence legible to the machine.
What Good Looks Like in Practice
For a service business or agency, a properly implemented homepage schema in 2026 should include a clean @graph with at least three connected nodes: Organisation (or LocalBusiness/ProfessionalService), WebSite, and WebPage. Each node should reference the others via @id, so Google can stitch the full entity picture together.
The Organisation node specifically needs: a legalName, address, telephone, email, geo coordinates, areaServed, sameAs pointing to every authoritative external profile, and an aggregateRating if you have reviews. The hasOfferCatalog property is worth including too, as it gives AI systems explicit signals about your service offering without relying on them to infer it from page content.
Blog posts need: Article or BlogPosting type, headline, datePublished, dateModified, author with a nested Person object that includes sameAs, and publisher pointing back to your Organisation @id.
None of this is particularly complex to implement. But it does require actually doing it rather than assuming a plugin has handled it.
The Bottom Line
Schema markup in 2026 is not a box-ticking exercise. It’s the infrastructure that connects your content to the AI-driven search ecosystem that an increasing proportion of your potential customers are using to make decisions. Getting it right doesn’t just help your rich results. It shapes how your brand is understood, cited, and recommended across every major search and AI platform.
The gap between businesses that have implemented structured data properly and those that haven’t is only going to grow as AI Overviews expand, and LLM-based search tools become more mainstream. The good news is that the technical barrier to getting it right isn’t that high. The barrier is mostly just knowing what you’re doing.
If you’re not sure where your schema currently stands, our free SEO audit will tell you exactly what’s there, what’s missing, and what needs fixing. No fluff, just a straight answer.
Jump into another related resource
Whether you’re after expert-written blogs, downloadable guides, or time-saving checklists, our Resource Hub gives you practical tools to make your marketing more effective.
