Does Schema Help AI Search? What Google Says and What Still Works

What does Google say about schema and AI features?

Google says schema is optional for its AI features. Its 2026 guide to generative AI search lists overfocusing on structured data among things site owners can ignore, adding that no special schema.org markup exists for AI and that markup still matters for rich results.

The source is Google Search Central’s Optimizing your website for generative AI features on Google Search, last updated 10 July 2026. It sits in a section headed “Mythbusting generative AI search”.

The wording is plain. Structured data “isn’t required for generative AI search”, and Google still calls it “a good idea” as part of overall SEO, because it helps with rich result eligibility.

A second page says the same thing from another angle. Google’s AI features and your website, updated 10 December 2025, lists the basics that help AI Overviews and AI Mode.

One item on that list is “making sure your structured data matches the visible text on the page”. Another says there is “no special schema.org structured data that you need to add”.

So Google treats schema as ordinary SEO hygiene. It is not a ticket into AI Overviews. A page still has to be indexed and eligible for a snippet, which is the actual entry rule Google gives.

That leaves a practical reading. If a vendor sells “AI schema” as the reason you will appear in AI Overviews, ask them to show the Google page that says so. There is none.

When Digital Magma reviews a schema proposal from another vendor, the first question is simple: which visible fact does each property describe? Markup that cannot answer that question gets cut.

What does schema markup still do in Google Search?

Schema markup still does three jobs Google documents: it makes pages eligible for rich results, helps Google tell your organization apart from others, and carries product and local details. None of these is an AI feature, but each feeds the same index AI Overviews draw on.

Google’s Introduction to structured data explains the first job. Structured data gives Google explicit clues about a page’s meaning and can make it eligible for richer search results.

Rich results

Product prices, review stars, event dates and recipe times all come from markup. These are visible results that can change click rates in normal search, and they need correct schema.

Entity clarity

Google’s Organization structured data page says the markup can help Google understand your organization’s administrative details and “disambiguate your organization in search results”.

That matters for AI answers in a roundabout way. A business that is easy to identify is easier to describe correctly, wherever the description appears.

Product and local data

Google’s AI guide says AI responses can include product listings and local business details. It points to Merchant Center feeds and Google Business Profile as the sources, not page schema.

Schema on your product and location pages should repeat those same facts. When the feed, the profile and the markup agree, there is less room for a wrong price or old address.

Does schema help in ChatGPT, Perplexity and Copilot search?

No primary documentation from OpenAI or Perplexity says their answer systems read schema markup. Microsoft’s advertising blog says schema helps search engines and AI systems understand content, without saying how Copilot weighs it. Treat any stronger claim as unproven.

Digital Magma checked the relevant pages from each company on 22 October 2026. This is what each one says:

The Microsoft source is a blog post, Optimizing Your Content for Inclusion in AI Search Answers. It says schema can label content as a product, review, FAQ or event so machines can interpret it.

That is the strongest primary statement in favour of schema for AI answers. It is still a general statement, with no ranking weight attached.

For ChatGPT, the honest position is “unknown”. OpenAI’s help article on searching the web says ChatGPT sometimes sends rewritten queries to search providers. If those providers use schema, it may matter one step removed.

Conference talks and vendor studies often claim more. A talk is not documentation, so Digital Magma does not repeat those claims to clients as fact.

Which schema types are worth having in 2026?

Five schema types earn their place on most business sites in 2026: Organization, LocalBusiness, Product, Article and BreadcrumbList. Each describes something real and visible on the page, and each has a Google documentation page explaining what it does in search.

1. Organization on the homepage or about page. Name, logo, URL, contact point and sameAs links to your official profiles.
2. LocalBusiness for each real location. Address, phone, opening hours and geo coordinates, matching your Google local business documentation and your Business Profile.
3. Product with Offer for anything you sell online. Price, availability and currency, matching your Merchant Center feed.
4. Article or BlogPosting on editorial content, with author and dates that match the byline.
5. BreadcrumbList on deep pages, so the site path is explicit.

Two more are worth knowing about, with care.

• FAQPage. Google said in August 2023 that FAQ rich results would only show for well-known, authoritative government and health websites. The markup is still valid schema.org, but most businesses will not see a rich result.
• Speakable. Google’s speakable documentation calls it a beta feature used by Google Assistant for topical news queries. It suits news publishers far more than service firms.

Digital Magma’s own site uses Organization, a founder Person, WebPage with author and reviewer, FAQPage from the visible FAQs, and DefinedTerm on knowledge-base entries. We keep it because it describes the site accurately, not because we expect AI rankings from it.

Which schema work wastes time in AI search?

Schema work wastes time when it marks up things the page does not show, chases types with no documented use, or is sold as an AI ranking trick. Google warns that structured data must match visible text, so invented markup is a risk as well as a waste.

These are the patterns Digital Magma removes most often during audits:

• Review stars for your own business. Self-serving review markup on a company’s own pages breaks Google’s review snippet rules and invites a manual action.
• Markup for hidden content. FAQ or HowTo markup for text that never appears on the page.
• QAPage on single-author pages. Google’s QAPage documentation is for pages where users can submit answers to one question. An expert article does not qualify.
• Stacking every type. A page marked as Service, Product, Article and FAQ at once tells no one what it is.
• “AI schema” plugins. Tools that promise special markup for ChatGPT or AI Overviews. Google says no such markup exists, and OpenAI documents none.
• Copied Wikidata IDs. A sameAs link to the wrong entity can merge you with another company. Check every ID before you use it.

HowTo needs a note too. Google announced in 2023 that HowTo rich results were limited to desktop, and its blog post now says they are no longer shown. The markup will not produce a visible result.

None of this means “remove all schema”. It means every property should describe something a visitor can see, and every type should have a reason to exist.

What Digital Magma has seen with schema on client sites

On the client sites where Digital Magma has seen AI visibility, schema was part of a sound build but never a lever we could isolate. The gains came with crawlable pages, direct answers, consistent business facts and search rankings, with schema quietly describing all of it.

Take the clearest AI-visibility case in our current work. A Canadian welding manufacturer, now three years into its engagement, holds 40+ keywords in Google’s top 10 and appears in ChatGPT and Google AI Overviews.

In one three-month span its site drew about 30K clicks and about 100K impressions. Its schema is correct, but we would not credit the markup with those numbers. We cannot separate its effect from everything else.

The UK tele-oncology startup we built from scratch tells a similar story. The decisive move was a React to Next.js migration that put content into the HTML. In month one, 30+ keywords ranked and 8 reached #1 on Google.

Markup could only describe content that crawlers could already read. That order of events shapes how Digital Magma prices and plans schema work today: it is a small, early task, not a campaign.

Where schema does show visible value for clients is in normal search: product rich results for shops and correct local details for clinics and service firms.

How to check your schema in 5 steps

A schema check takes under an hour for most small sites. Digital Magma runs these five steps in order: list what exists, validate it, compare it with the visible page, compare it with your profiles and feeds, then remove anything that fails.

1. List what you have. Run your homepage and three key pages through the Rich Results Test. Note every type it finds.
2. Validate the syntax. Use the Schema Markup Validator for types Google does not show as rich results.
3. Match the visible page. For each property, find the same fact on screen. Price, phone, hours, author and dates must agree.
4. Match your profiles. Compare Organization and LocalBusiness data with your Google Business Profile and any Merchant Center feed.
5. Remove what fails. Delete markup for hidden content, self-serving reviews and types with no reason to be there.

Digital Magma logs each schema finding in a short sheet: page, type, property, problem and fix. That record makes the next audit faster.

Then check Search Console’s enhancement reports a week later. Any new error there means the markup and the page have drifted apart.

Repeat the check after a theme change, a plugin update or a price change. Those are the moments schema most often breaks without anyone noticing.

Schema and AI search: frequently asked questions

These are the schema questions clients bring to Digital Magma when AI search comes up. The answers stick to what Google, Microsoft and schema.org document, and say plainly where documentation stops.

Related schema questions, sources and author

These pages cover the schema topics around this answer. Digital Magma marks planned entries until they are published, and links live knowledge-base terms directly.

• Google AI Overviews: how Google’s AI summaries pick supporting links
• Answer engine optimization: writing pages that answer first
• llms.txt: another machine-readable file Google says it ignores
• Schema markup, entity SEO and Knowledge Graph (planned entries)
• What makes content easy for AI to quote? and Should you block AI crawlers?

Sources: Google Search Central’s AI optimization guide, AI features page, structured data introduction, Organization, LocalBusiness, QAPage and speakable documentation, its August 2023 blog post on FAQ and HowTo changes, and Microsoft Advertising’s October 2025 blog post. All were checked on 22 October 2026.