Home › AI Search › Schema and structured data
Schema and Structured Data for AI Search
Structured data does not make an AI cite you, and Google says so directly. What it does is remove the ambiguity that stops one. Schema states what a page is, who wrote it, when it was updated, what your organisation is called and where it operates, in a form nothing has to infer from prose. Five types carry almost all of the value for a service business: Organization, LocalBusiness, Service, Article with a named author, and FAQPage. Everything after that is refinement.
Start with what Google actually says
This is oversold more consistently than anything else in the field, so it is worth quoting the source rather than the commentary. Google's guide to optimising for generative AI features states that structured data is not required for generative AI search and that there is no special schema markup you need to add for it. Google's general structured data guidelines separately state that Google does not guarantee your structured data will show up in search results even if the page is marked up correctly.
So if you have been sold schema as an AI ranking factor, you have been sold something Google has explicitly denied. That does not make it worthless. It makes the reason for doing it different from the reason you were given.
The real reasons to do it
- Rich results in ordinary search, which is where nearly all of your traffic still comes from. This is the direct commercial return.
- Entity clarity. Schema states your name, address, phone number and identity as data rather than as prose, so nothing has to guess which business you are.
- Authorship. Article markup with a named author, a datePublished and a dateModified is how a page asserts who stands behind it and when it was current.
- Consistency. Marked up facts are much easier to keep identical across a site than facts written into sentences by different people over several years.
The five types that carry the value
| Type | What it settles | Where it goes |
|---|---|---|
| Organization | Who you are, with links out to the profiles that corroborate it | Site wide |
| LocalBusiness | Where you are and where you serve | Home page and location pages |
| Service | What a given service is and who it is for | Each service page |
| Article | Who wrote this, when, and when it was last updated | Guides, insights and definitional pages |
| FAQPage | A question and its answer, as a pair | Only where the questions are genuinely unique to that page |
Resist the urge to add more. A page carrying six overlapping types is not six times as clear, and conflicting markup is worse than none. Check Google's own list of supported types rather than a copied list in a blog post, because the set changes.
The two fields most sites leave out
A named author and an honest dateModified. Most sites either omit the author entirely or attribute everything to a generic account, and most set dateModified automatically so that it changes when a plugin updates rather than when the content does. Both undermine the point of having the markup at all. If your dateModified says last week and the page has not been touched in two years, you have automated a small lie.
How to check it, and how not to
Use Google's Rich Results Test or the Schema Markup Validator on the live URL. Do not check structured data with a text extraction tool or a content scraper, because those strip script tags before you see the output. JSON-LD lives in a script tag, so it becomes invisible and every site you check appears to have no structured data at all.
That mistake cost us most of a day on a competitor audit earlier this year, and it is worth naming because a great many automated AI readiness reports are built on exactly that method. If a report tells you a well built site has no structured data, check the raw HTML before you believe it.
If you are on WordPress with Rank Math
- Rank Math already outputs a full graph on most pages, typically Organization, WebSite, BreadcrumbList, WebPage, Person and Article. Look at what it produces before you add anything, or you will duplicate it.
- FAQ markup needs the FAQ block to be present in the page content itself. Configuring it only in the Schema Generator will validate in the plugin and render nothing on the page.
- Check that the Article author resolves to a real user with a real profile, not to the account somebody used to build the site.
- After any change, view the rendered page rather than the database, and purge your cache first. Cached HTML hides new markup and has fooled better people than us.
What to do if you have none of this
Start with Organization and LocalBusiness, get your name, address and phone number identical everywhere they appear, and stop. That single pass does more for entity clarity than any amount of elaborate markup on individual pages, and it is usually an afternoon's work rather than a project.
Read next
Frequently asked questions
Does adding schema improve our Google rankings?
Not directly. Google's guidelines say structured data is not a guarantee of anything appearing, and its generative AI guidance says no special markup is needed for AI features. The gain is eligibility for rich results in ordinary search, plus clearer entity signals.
Is FAQ schema still worth adding?
Where the questions are genuinely unique to that page, yes. Where the same block is repeated across many pages to inflate coverage, no. That is the pattern Google's scaled content abuse policy is aimed at, and it also makes your pages compete with each other.
Can we add structured data without a developer?
On WordPress, usually yes. Rank Math and Yoast both generate the core types from settings, and the main work is filling in the organisation details accurately and checking the author is a real person. Custom types are where a developer becomes worthwhile.
Why does our schema checker report that we have none?
Almost always because the checker extracts text rather than parsing raw HTML, which strips the script tag that JSON-LD lives in. Test the live URL in Google's Rich Results Test instead, and compare the two before concluding anything.
Not sure how AI assistants describe your business?
We run a hand checked AI visibility audit, scored out of 100 across five areas, with the weightings published so you can argue with them.
Book a 15 minute callSources
- Google Search Central. Google's Guide to Optimizing for Generative AI Features on Google Search. Updated 10 July 2026.Generative features are rooted in core ranking systems. No special schema required. Google Search ignores llms.txt.
- Google Search Central. General Structured Data Guidelines. Updated 10 July 2026.Google does not guarantee that structured data will show in results even if marked up correctly.
- Google Search Central. Structured data markup that Google Search supports. Updated 15 June 2026.The current list of supported structured data types. Check it rather than trusting a copied list.
- Google Search Central. Spam policies for Google web search. Updated 15 May 2026.Scaled content abuse and doorway abuse, in Google's own wording.
Every figure on this page was read on the source page listed above. This page is reviewed monthly and the last checked date is updated when it is.
