Schema markup for AI: what it helps and what it cannot do

Schema markup is a set of labels hidden in your website’s code. They tell search engines facts such as your business name, your web address and the suburbs you serve. Picture a pantry shelf of jars, where the label on each tells you what’s inside without opening it. A label that doesn’t match the jar misleads everyone who trusts it.

Here’s an example. Say a web developer or a plugin salesperson tells you that you need schema markup to get found by ChatGPT. Before you pay for it, you should know what Google has said, what the labels can honestly do and how to tell whether they helped.

The short version: schema markup is a clear description of your page for machines and it is worth getting right. It isn’t a switch that puts a business into AI answers. Our AI SEO agency checks the markup a site already has and fixes anything that contradicts the page. We then measure the results one step at a time. Work on how a business shows up in AI answers is called generative engine optimisation (GEO). Our guide on what is GEO covers it more widely.

What is schema markup for AI search?

Schema markup is code that describes a page’s facts in a format machines can read. Google calls these standardised page descriptions structured data and uses them to understand content.¹

Three words to know. Schema.org is the shared vocabulary for describing businesses, people and articles. Structured data is a description written with that vocabulary. JSON-LD is one format for placing it in a page and Google recommends it where your setup allows.¹

Google doesn’t require these descriptions for its AI-written search answers. It also doesn’t need a special version for AI Overviews and AI Mode, the two places Google shows them.² That’s what Google says about its own products. The sources we read don’t show how other assistants, such as ChatGPT, use a page’s descriptions.

Here’s what the code looks like for a fictional tiling business, so the business and web addresses are placeholders.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Harbour Tiling Co",
  "url": "https://www.example.com.au",
  "sameAs": ["https://www.facebook.com/example-harbour-tiling"],
  "areaServed": "Inner West, Sydney"
}

Every fact in it is something a customer could also read on the page, which is the test to keep in mind. The picture below shows three of those facts side by side. Each one is visible on the page and has a matching line in the code.

  "name": "Harbour Tiling Co",

  "url": "https://www.example.com.au",

  "areaServed": "Inner West, Sydney"

Does schema markup make a page appear in AI answers?

Not by itself, because Google doesn’t require special schema markup for its AI-written answers. A page is eligible when Google has stored it for search, which is called indexing. The page must also be able to show a short excerpt and be included in Google’s AI search features. Even then, Google doesn’t guarantee that an eligible page will be shown.²

Markup does count in ordinary Google search. Google says structured data helps with eligibility for rich results, which are enhanced listings with extra detail.² Even there, a page can miss out if it ignores Google’s rules for the data type or leaves out required fields, which Google calls properties.¹ We read that as a chance and not a guarantee, so valid markup alone doesn’t promise a rich result.

Say someone offers to add schema to your site. Ask which visible facts the code will describe, which search feature it supports and how they will check it. If they say a change won you more citations, ask for the saved answers and a list of the other changes made at the same time. A citation is a link an AI answer shows as its source.

Kavera note: In our Sydney trades AI search study we asked 50 questions across four AI answer products and recorded which businesses and links came back. We didn’t test markup in that study, so it can’t say whether schema changed any answer. We use it to show how to measure and we don’t offer it as schema evidence.

Which schema types fit a local service business?

Choose the description that matches what is on the page. These four Schema.org types are starting points for business details, articles and questions, so use only the ones that fit the page.

LocalBusiness belongs to both Organization and Place in Schema.org’s tree, so it has every field an Organization has.⁶ It adds fields such as openingHours and priceRange.⁶

Say you’re a tiler who travels to customers and has no shopfront. An Organization with an areaServed value may describe you more honestly than a LocalBusiness with an invented address. The areaServed field is the area where a service or offered item is provided.⁵

Google bans markup that is misleading or irrelevant, including fake review claims.³ Genuine reviews that visitors can see on the page are only a starting condition. Before you add review markup, check Google’s requirements for that page type and search feature.

Which business details should you keep consistent?

Keep your business name, web address, service area, logo and profile links the same everywhere they appear. The aim is one master copy of each fact, so your visible text and your markup both come from it.

Google’s rules sit behind that table. Structured data must be a true representation of the page content. You must not mark up content that readers of the page can’t see.³ A breach can bring a manual action, which means the page loses eligibility for rich results. Google says rankings in standard web results aren’t affected by that action.³

Use sameAs only for a page that clearly identifies the same business. That can be an official profile or a reference such as its Wikipedia or Wikidata page.⁵ A well-known site that has nothing to do with your business doesn’t become a reference for it just because you link to it.

Is FAQ schema still useful?

Not in the way it used to be. Google stopped showing expanded question-and-answer listings, called FAQ rich results, from 7 May 2026.⁴ It removed the feature’s documentation on 15 June 2026.⁴ Schema.org still defines FAQPage for describing a visible page of questions and answers.⁷ Keeping that description doesn’t restore the retired Google display.

That doesn’t make a visible FAQ pointless, because customers still ask the same questions before they call. A short list of honest answers on a service page helps a reader decide. AI tools can also quote a clear answer. Keep the questions you hear most and leave out the ones you wrote only for markup.

Kavera note: On 20 October 2026 we read the live what is GEO guide. It showed a visible FAQ and machine-readable page descriptions, called structured data, using Organization, BlogPosting, BreadcrumbList and FAQPage. We keep the questions for readers without expecting the retired expanded Google listing.

How do you check the markup you already have?

Look at the code your website already outputs before you add anything. Here’s an example. Say your theme and an SEO plugin both describe your business with different opening hours. Compare their details and fix the conflict at the source.

1. Test one key page with the Rich Results Test, which Google calls an easy and useful tool for validating structured data.¹ The Schema Markup Validator is a second tool for checking Schema.org markup.⁹
2. Read every field and compare the business name, address, hours, service area and any prices with what a visitor sees.
3. Look for duplicates by checking whether two tools describe the same business or article, then decide which one is the master copy.
4. Fix it where it starts. Change the plugin setting or template that creates the markup instead of pasting a second block over the top.
5. Check again after launch. Google advises using rich result status reports after you deploy, to catch problems introduced by templates or by the way pages are served.¹

Start with your home page, your main service pages and one article. Record what you checked and which template supplies each description before you expand the work across the site.

How can you tell whether schema work helped?

Check three things: whether the code passes the testing tools, whether any enhanced listings appear and whether AI answers mention your business. Then count visits and enquiries on their own. Passing a code check doesn’t prove any of the later results.

Keep a dated log of every change you make, including content edits, profile fixes and markup. If three things changed in one month, you can’t credit any one of them. Our guide to measuring AI visibility shows a monthly routine for the saved-answers row.

When is it worth paying for schema work?

Pay for it when your existing markup is wrong, duplicated or missing for a feature you want. Make sure someone checks the result against the page. Don’t pay for it as a route into AI answers.

Ask any provider four questions:

1. Which types will you add and which facts will each describe?
2. Where does the data come from, so the page and the markup stay in step?
3. How will you test it?
4. What will you measure afterwards?

A good answer names a validation tool and a Search Console report while a poor answer promises citations.

The code can be correct and still do nothing if search tools can’t retrieve the page. Our guide to AI crawlers and robots.txt explains the automatic page readers and permission file involved.

What should you do first?

Test your home page and one service page in the Rich Results Test this week. Compare what the tool found with what a visitor sees, fix any contradiction at its source and write down what you changed.

If you want a second opinion, our AI SEO agency can review your markup with you. Our SEO audit is free. It reviews your technical setup, keywords, competitors and Google Business Profile, then gives you a prioritised roadmap. Implementation is separate unless retained. Monthly plans start at Suburb SEO, $890/month + GST. See the SEO packages for scope and terms. No SEO package guarantees rankings, leads or revenue.