Technical AI SEO

Schema for AI Search: What Actually Helps in 2026?

The schema for AI search should describe visible content accurately; it is not a special code that guarantees citations. Google says AI Overviews and AI Mode require no AI-specific schema. Use precise Organization (sic) or LocalBusiness, Service, BlogPosting, BreadcrumbList and genuine FAQPage markup to reduce ambiguity, connect entities and support standard search eligibility.

Structured data is valuable when it makes the page more explicit. It becomes risky when the markup claims services, ratings, locations or answers that visitors cannot verify in the visible content.

Learn how we deploy structured entity graphs across client sites in our core AI Search Optimisation services, as well as how technical schema markup forms Pillar 1 of our local SEO Wollongong framework for regional businesses. Short on time? Watch the 2-minute video.

Schema · AEO · GEO · Updated November 2026

Watch: Schema for AI Search in 2 Minutes

Simon Bayliss explains what structured data can and can't do for AI search, the five schema types worth using, and a simple test for whether markup belongs on a page.

Is a special AI schema required?

No. Google states that there are no additional technical requirements for AI Overviews or AI Mode and no special schema.org markup to add. A supporting page must be indexed, eligible for a snippet and compliant with normal search requirements.¹

This distinction matters. Schema is a standardised description layer; it is not an “AI Overview switch”. Google still recommends useful content, crawl access, internal links, text availability, good page experience, current business profile information and structured data that matches visible text.

Evaluate whether your current structured data passes LLM parsing with a fixed-price AI SEO Audit

Creative Orbit framework — Visible Truth → Structured Identity → External Proof: state the fact clearly on the page, encode the same fact consistently in structured data, then corroborate important claims through reviews, citations, profiles and credible third-party sources.

Which schema types are useful?

Choose the most specific type that reflects the page’s main purpose. Google’s guidelines require structured data to be a true representation of visible content and recommend placing markup on the page it describes.²

Organisation or LocalBusiness

Define the business identity, URL, contact details, address or service area, and trusted sameAs profiles. Use a specific subtype where it genuinely applies.

Service

Describe one visible service, its provider and the area served. Do not generate a catalogue of services that the page never discusses.

BlogPosting

Identify the article headline, author, publisher, dates, image and canonical page so ownership and freshness are explicit.

BreadcrumbList

Express the page’s place in the site hierarchy and align the markup with visible, crawlable navigation.

FAQPage

Mark up genuine questions and answers visible on the page. Keep the schema text aligned exactly with the rendered FAQ content.

JSON-LD is Google’s recommended format in most cases because it separates the data block from visible markup and is generally easier to maintain.³

The three-question schema test

Before adding any markup, answer three questions:

What is this page for? Name its main commercial or informational purpose.
What can a visitor actually see? Every claim, price, location or FAQ answer in the schema must be visible in the page content.
Which schema type matches that purpose? Choose the type that fits, and don't stretch markup into categories the page doesn't support.

What changed for the FAQ schema?

Google stopped showing the FAQ rich-result feature in search from 7 May 2026.⁴ That change removes the old visual incentive; it does not make clear, useful on-page FAQs irrelevant.

Keep FAQPage markup where the page genuinely contains an FAQ and the structured answer matches what users can read. Remove low-value boilerplate written only to occupy search-result space. The strategic purpose is now semantic clarity and answer usability, not an expandable rich-result promise.

Use FAQ markup when

The questions are specific to the page’s topic and actual audience needs.
Each answer is visible, self-contained, factual and current.
The JSON-LD matches the on-page wording.
The section improves the page even if no search feature appears.

How should entities connect?

Use stable identifiers and consistent relationships so the same organisation, person, service and webpage are not described as disconnected objects.

Give the organisation a stable @id, usually based on its canonical website URL plus a fragment.
Reference that same entity as the publisher, provider or parent organisation where appropriate.
Use canonical URLs for the webpage and article.
Link the named author to a substantive profile page.
Use sameAs only for authoritative profiles representing the same entity.
Keep names, URLs, service areas and contact details consistent with visible content.

For Creative Orbit, that means a clear connection between Simon Bayliss, the Creative Orbit organisation, the Wollongong service location, the SEO Wollongong service and the AI search optimisation service. The graph should clarify real relationships, not manufacture authority.

Want to know if AI search recommends you? Get an AI Search Visibility Audit →

What technical checks matter?

Valid schema cannot rescue a page that crawlers cannot access or users cannot read. Test the full delivery path, not only the JSON-LD syntax.

Run this deployment checklist

Confirm the page returns a successful response and is indexable.
Ensure important content is present in the rendered and source HTML.
Validate schema syntax and the requirements of any targeted Google feature.
Compare every marked-up claim with the visible page.
Check canonical, dates, author, image URLs and breadcrumb destinations.
Allow the search crawlers needed for the platforms you want to reach.
Monitor Search Console and server logs after deployment.

OpenAI says OAI-SearchBot is used to surface websites in ChatGPT search and that sites opting it out will not be shown in ChatGPT search answers, although navigational links may still appear.⁵ Crawler access and schema solve different problems: access enables retrieval; structured data improves explicitness.

Does llms.txt replace schema?

No. Google explicitly says llms.txt and other special AI text files are not required for Google Search or its generative AI capabilities.⁶

An llms.txt experiment may be maintained for systems that choose to use it, but it should not compete with the essentials: crawlable server-delivered content, accurate canonical pages, internal links, valid standard schema and a coherent external entity footprint. Treat it as optional documentation, not a substitute for technical SEO. How the website platform shapes what crawlers receive is covered in WordPress Alternatives for 2027.

A page-specific SEO and AI-readiness audit should therefore test retrieval, rendering, structured data, entity consistency and content evidence as one system.

Video transcript and key moments ▼

00:00 Schema for AI Search: The Real Question: "Schema for AI search is often presented as a technical shortcut. But what does the evidence actually support?"

00:07 What Schema Can and Cannot Do: "Understanding the clear boundaries of structured data is essential. It provides machine-readable disambiguation, but it cannot compensate for low-quality content or unverified entities."

00:15 The Central Principle: Describe Visible Content: "Schema must accurately describe what human users see on the screen. Injecting invisible facts or unverified assertions into JSON-LD creates semantic conflicts."

00:25 No Special Schema for AI Overviews or AI Mode: "Google's own guidance confirms that AI features do not rely on exclusive, secret tags. They parse standard, high-quality on-page content and established schema definitions."

00:34 Make the Page and Business Clear: "Instead of searching for novel hacks, focus on making your business identity, physical footprint, and core services completely unambiguous to indexing models."

00:43 Schema Must Match Visible Content: "Relevance and strict factual accuracy matter far more than novelty. If the visitor cannot read it in the text, it does not belong in your schema graph."

00:56 Five Practical Schema Types: "Rather than cluttering pages with dozens of obscure schemas, focus your implementation on five high-value, evidence-backed types."

01:10 Organisation and LocalBusiness Schema: "Establish corporate entity ground truth: legal names, Australian Business Numbers (ABN), physical addresses, telephone points, and explicit service areas." See how this integrates with local map visibility on our What Local SEO Includes Briefing.

01:17 Service Schema for Genuine Offerings: "Apply the service schema directly to dedicated service landing pages to define clear capabilities, deliverables, and commercial scopes."

01:25 BlogPosting and BreadcrumbList: "Use BlogPosting on research papers and articles to attribute recognised human authors and timestamps. Deploy BreadcrumbList across every template to make the site hierarchy machine-navigable."

01:40 FAQPage: When It Is Appropriate: "Only mark up genuine, visible question-and-answer content. Never inject speculative FAQ blocks purely in the hope of earning rich snippets or AI citations."

01:54 The Practical Schema Test: "Before publishing, ask: What is this page for? What can visitors see? Which standard schema type most accurately represents that purpose?"

02:05 Validate and Maintain Your Markup: "Structured data is not a set-and-forget task. Whenever on-page copy, pricing, or service details are updated, your JSON-LD graphs must be updated and validated to match."

02:11 Schema Is Not a Shortcut: "Schema cannot rescue thin, unhelpful content. It serves solely to provide a verified, unambiguous summary of genuine expertise."

02:17 Evidence-Led SEO and AI Search Strategy: "Creative Orbit implements evidence-based structured data graphs designed to withstand core algorithm updates and build long-term AI answer engine authority."

Frequently asked questions

Is there a special schema for Google AI Overviews?

No. Google states that no special schema.org structured data is required for AI Overviews or AI Mode. Pages still need standard crawlability, indexability, snippet eligibility, helpful content and structured data that matches the visible page.

Does schema guarantee an AI citation?

No. Structured data can reduce ambiguity and help machines understand page entities, but it does not guarantee selection, ranking or citation. Content quality, technical eligibility, authority, corroboration and the needs of each generated answer still matter.

Should FAQPage schema still be used in 2026?

FAQPage remains a valid Schema.org type, but Google stopped showing its FAQ rich-result feature in Search from 7 May 2026. Use FAQ markup only when the page contains genuine visible questions and matching answers, not to chase a discontinued visual result.

Which schema types suit a local service business?

Use the most specific valid types that match the page: Organisation or LocalBusiness for the entity, Service for an actual service, BlogPosting for an article, BreadcrumbList for navigation and FAQPage for a genuine FAQ section. Required and recommended properties depend on the type and search feature.

Does Google use llms.txt for AI search?

Google says llms.txt and other special AI text files are not needed for Google Search or its generative AI features. Other systems may handle such files differently, but llms.txt is not a substitute for crawlable HTML, internal links, accurate content or standard technical SEO.

co-mod-cta-box-ai

References