AI Overview Markup: What Google Actually Requires (and What It Ignores)

Google does not require a special AI Overview tag, schema type, Markdown file or llms.txt file. Its current guidance says the same SEO foundations used for conventional Search also apply to AI Overviews and AI Mode.

That does not mean businesses should do nothing. It means the commercial opportunity is in improving the substance, structure and discoverability of the pages buyers already need, instead of paying for unsupported shortcuts.

John Russell, Director of Digital Womble: “There is no special AI Overview markup. The advantage comes from doing the fundamentals better: useful content, clear structure, strong internal links and accurate data.”

Reviewed against current Google Search Central documentation on 10 September 2026.

The Short Answer

If you want a page to be eligible as a supporting link in Google AI Overviews, focus on five things:

• Publish unique, useful content grounded in genuine experience or expertise.
• Make the page crawlable, indexable and eligible to appear with a search snippet.
• Use clear titles, headings, link text and descriptive image alt text.
• Add relevant internal links that Google and visitors can follow.
• Keep ordinary structured data accurate and consistent with the visible page.

Google states that there are “no additional technical requirements” for appearing as a supporting link in AI features. It also says there is no special schema.org markup required for generative AI search.

Eligibility is not a guarantee. A page can meet every technical requirement and still not be crawled, indexed, ranked or selected in an AI-generated answer.

What Google Says You Do Not Need

The market has created a new vocabulary around AI search. Some of it is useful. Some of it simply repackages SEO fundamentals as a new technical product.

According to Google’s guidance for generative AI features, you do not need:

• A special AI Overview schema type
• New “AI text” files or machine-readable markup
• An llms.txt file for Google Search
• Artificially tiny content chunks
• A separate version of an article written only for AI systems
• Every possible long-tail variation forced onto its own page
• Manufactured mentions intended to make a brand look popular

An llms.txt file may be useful to another service that supports it. Google says it neither helps nor harms visibility in Google Search because Google Search ignores it.

Structured data still has value, but for its normal purpose. Valid Article or Product structured data can help Google understand a page and make it eligible for relevant rich results. It is not an AI Overview admission ticket.

What Actually Matters for AI Overview Visibility

1. Information worth retrieving

Google’s current guidance puts unique, compelling and useful content ahead of technical tricks. A page that merely restates the same generic answer available across dozens of websites gives an AI system little reason to retrieve it.

Original value can come from:

• First-hand experience
• Proprietary data
• A practical method
• A clearly expressed expert viewpoint
• Real examples and outcomes
• Product specifications or operational knowledge
• A better answer to a real customer question

This is where content and revenue strategy meet. A page should not exist merely because a keyword has volume. It should help a likely customer make a decision that matters to the business.

If you do not know which questions connect to demand, build a commercially focused SEO content strategy before producing another batch of isolated articles.

2. Clear titles, headings and answer structure

Google’s Search Essentials recommends using the words people use to find information in prominent places such as the page title, main heading, alt text and link text.

Clear structure benefits three audiences at once:

• Readers can understand the answer quickly.
• Search systems can identify the page’s subject and relevant passages.
• Your team can connect each section to a useful next step.

This does not require writing robotic two-sentence “chunks”. Google explicitly says there is no required page length or chunk size for AI search. Use the length and structure the subject deserves.

A useful commercial page often works well when it includes:

• A direct answer near the top
• Descriptive H2 headings
• Short paragraphs
• Bullets for genuine lists
• Evidence beside the claim it supports
• A clear conclusion and next action
• FAQs only where they answer real objections

3. Crawlable internal links

Internal links help visitors continue their journey and help Google find and understand related pages. They should use ordinary crawlable links with descriptive anchor text.

For example, a reader who understands the basics may need the more detailed Google AI Overview SEO technical checklist. A business that is unsure whether it appears in AI answers may be better served by an AI visibility benchmark.

The anchor should explain what the destination provides. Avoid “click here”, broken links and links inserted solely because a phrase happens to match.

Commercially, this is crucial. Informational content creates value only when the right reader can move naturally towards a service, product, trial, assessment or deeper buying-stage page.

4. Useful images with meaningful metadata

Google recommends high-quality images near relevant text. It says filenames can provide light clues about the image subject, while descriptive alt text is the most important image attribute for providing additional context and improving accessibility.

A sensible image workflow therefore includes:

• A relevant image rather than decorative filler
• A short, descriptive filename
• Contextual alt text that describes the image
• A stable, crawlable image URL
• Width and height information to reduce layout movement
• An image reference in structured data where the chosen rich-result type requires it

Alt text is not a place to repeat a target keyword. Describe what is actually shown and connect that description naturally to the page context.

5. Accurate structured data

Google says structured data should match the visible page. For a blog article, Article or BlogPosting structured data can help Google understand information such as the title, author, image and publication date.

Use the most specific appropriate type, include only facts that are visible or verifiable, and validate the result.

Do not:

• Add unsupported claims to the markup
• Mark up an FAQ that users cannot see
• Invent reviews or ratings
• Treat schema volume as a substitute for useful content
• Claim that structured data guarantees an AI citation

The right principle is simple: structured data should accurately describe the page you built for people.

The Revenue Test Most AI Overview Advice Misses

Visibility alone is not the commercial goal.

A page can gain impressions, rankings or citations and still produce no useful business result. Before creating it, decide:

• Which buyer is this page for?
• Which commercial problem does it support?
• What should the reader understand or do next?
• Which product or service page should it support?
• How will you distinguish useful demand from irrelevant traffic?

For a service business, that may mean leading an informed reader towards an assessment or consultation. For an ecommerce business, it may mean connecting a comparison question to an appropriate category or product. For a B2B supplier, it may mean turning technical specifications and customer objections into decision-support content.

Digital Womble’s approach starts with the commercial objective, not the writing prompt. The Content Strategy platform turns one important subject into a connected pillar and supporting articles, with internal-link recommendations, verified sources where applicable, relevant images and export-ready content.

A Practical AI Overview Content Checklist

Use this before approving a new or revised page.

Purpose and originality

• Does the page answer a real customer question?
• Does it add experience, evidence, data or a useful point of view?
• Is its intended reader clear?
• Is there a commercially appropriate next step?
• Does it avoid repeating an existing page with the same intent?

On-page clarity

• Does the title accurately describe the page?
• Is there one clear main heading?
• Is the direct answer easy to find?
• Do subheadings explain the sections beneath them?
• Are claims supported near the point where they appear?

Discovery

• Is the URL crawlable and indexable?
• Does another relevant page link to it?
• Are the links standard HTML links?
• Do the anchor words describe the destinations?
• Does the canonical URL point to the correct version?

Images

• Is each important image relevant to the surrounding text?
• Is the filename short and descriptive?
• Does the alt text describe the image accurately?
• Is the image URL stable and accessible to crawlers?
• Does the page reserve space for the image?

Structured data

• Is the selected schema type appropriate?
• Does every marked-up fact match the visible page?
• Are author, publisher, dates and image references accurate?
• Has the markup been validated?
• Are you using it for a supported rich-result purpose, rather than calling it special AI markup?

Measurement

• Is the page included in Search Console monitoring?
• Are conversions or meaningful next actions tracked?
• Can you distinguish discovery from commercial engagement?
• Will the content be reviewed when products, evidence or Google guidance change?

What Should You Fix First?

If the site cannot be crawled or the page is excluded from the index, fix that first.

If the page is technically eligible but says nothing distinctive, improve the content.

If useful articles sit in isolation, strengthen internal links and the surrounding topic coverage.

If visitors arrive but do not progress, improve the offer, proof, calls to action and destination pages.

If you do not know where the problem is, start with Digital Womble’s free three-page Search Revenue Diagnosis. It is designed to identify the most material opportunity before you pay for unnecessary work.

If the gap is a lack of commercially focused content, build a complete SEO Content Strategy for £48. One brief produces a pillar article and twelve supporting articles ready to review, edit and export. The tool supports the workflow, but human review remains essential and no ranking, traffic, indexing, citation or revenue outcome is guaranteed.

Frequently Asked Questions

Do I need special schema for Google AI Overviews?

No. Google says there is no special schema.org markup required for generative AI search. Continue using relevant structured data for its established Search and rich-result purposes, and make sure it matches the visible page.

Does Google use llms.txt for AI Overviews?

No. Google’s current guidance says Google Search ignores llms.txt and similar AI text files. You can maintain one for another service that supports it, but it does not improve Google Search visibility.

Should content be broken into tiny chunks for AI systems?

No fixed chunk size is required. Clear headings, readable paragraphs and useful lists help people navigate a page, but Google advises creating the page for the audience rather than forcing an AI-only format.

Can structured data guarantee an AI Overview citation?

No. Structured data can help Google understand a page and support eligibility for certain rich results. It does not guarantee crawling, indexing, ranking or selection as an AI Overview source.

What is the fastest useful first step?

Check whether the page is crawlable and indexed, then compare its answer with the strongest existing sources. If it is eligible but generic, add genuine experience, evidence and a clearer customer next step. If the wider content plan is unclear, create a connected strategy before scaling production.

Sources

• Google Search Central: Optimising your website for generative AI features
• Google Search Central: AI features and your website
• Google Search Essentials
• Google image SEO best practices
• Google Article structured data guidance