AI Visibility · 19 July 2026

By Eddie Eastwood, Founder

What Google Actually Says About Optimising for Its AI

Two checklists that cancel each other out

Picture an owner with two browser tabs open. One "AI SEO" checklist tells them to add an llms.txt file, chop every page into tiny chunks, and rewrite their copy in a special machine style. The other tells them none of that matters and only "mentions" count. The two lists contradict each other on almost every line, and both are written by people selling something.

There is a simpler move than picking a side: read what the platform itself publishes. Google maintains a public guide for site owners on how to appear in its generative AI features, and it is refreshed as those features change. Instead of trusting a third party's interpretation, you can quote the source.

Google published its own guide, so start there

Google's Search Central documentation now carries a page titled "Optimising your website for generative AI features on Google Search", its guide to AI Overviews and AI Mode. A companion page, "AI features and your website", sits alongside it. Both are official Google documentation, updated in 2026, and both are free to read. Everything in this article is drawn from those pages and the linked structured-data and helpful-content docs, quoted directly.

That matters because the guide contradicts a lot of the advice being sold around it. When a tactic and the platform's own documentation disagree, the documentation is the safer bet.

"Is SEO still relevant for generative AI search?" Google answers its own question

The guide opens by asking exactly the question owners are anxious about, and answers it plainly:

"In short, yes! The best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems."^[1]

It goes further on the "AEO" and "GEO" labels being marketed as a new discipline:

"From Google Search's perspective, optimising for generative AI search is optimising for the search experience, and thus still SEO."^[1]

In other words, Google frames its AI features as a new surface on top of the same ranking machinery, not a separate game with separate rules. If you want the background on those labels, our explainer on what answer engine optimisation actually is covers the same ground from the outside.

The one technical rule for appearing in AI Overviews and AI Mode

Owners expect a long technical checklist for the AI surfaces. Google's companion page says the opposite. To be eligible as a supporting link in AI Overviews or AI Mode, a page needs to be indexed and eligible to show in ordinary Search with a snippet, and then:

"There are no additional technical requirements."^[2]

The same page adds that the wider fundamentals still pull their weight:

"While specific optimisation isn't required for AI Overviews and AI Mode, all existing SEO fundamentals continue to be worthwhile".^[2]

So the entry ticket to the AI features is the entry ticket to Search: get indexed, be crawlable, be eligible for a snippet. There is no secret second gate. Our walkthrough of how AI Overviews work and how to show up in them unpacks what "eligible" looks like in practice.

Unique, non-commodity content beats every trick

If the guide has a single load-bearing sentence, it is this one, and it outranks every technical suggestion Google lists:

"Creating content that people find unique, compelling, and useful will likely influence your website's presence in generative AI search in the long run more than any of the other suggestions in this guide."^[1]

Google draws a sharp line between "commodity" content (its own example is a generic "7 Tips for First-Time Homebuyers" post) and "non-commodity" content that carries a first-hand or expert take the reader could not get from a summary. For a local business that is good news: you have first-hand experience of the jobs you actually do, and a content mill does not. Writing from that is what Google is asking for.

This is the same principle behind how AI engines pick which sources to trust: first-hand substance travels further than volume.

The same fundamentals, mapped

Google's guide reframes familiar SEO fundamentals for the AI context rather than replacing them. Read plainly, the chain is short: make content people-first and genuinely useful, keep the technical structure clean so it can be crawled and indexed, and keep your local and product details current so they can be surfaced. The guide's own framing of people-first content comes from its helpful-content page:

"People-first content means content that's created primarily for people, and not to manipulate search engine rankings."^[3]

Nothing in that chain is AI-specific. It is the ordinary work, done properly.

What Google says you can ignore

The guide has a section headed "Mythbusting generative AI search: what you don't need to do". It is the part most worth reading. It names the exact tactics being sold as essential and tells you to skip them for Google Search. Each one is quoted against the hack it answers in the table below.

Google's guide does note that creating an llms.txt file will neither help nor harm your Google ranking, so it is not forbidden, just ignored by Google Search.

Where structured data still earns its place

"Not required for AI" is not the same as "pointless". Google's guide recommends keeping structured data as part of your overall strategy because it helps a page qualify for rich results in ordinary Search. Google's own structured-data documentation carries the case-study numbers, and they are worth keeping in proportion:

• Nestlé measured that "pages that show as rich results in search have an 82% higher click through rate than non-rich result pages."^[4]
• Rotten Tomatoes "added structured data to 100,000 unique pages and measured a 25% higher click-through rate for pages enhanced with structured data, compared to pages without structured data."^[4]
• The Food Network "has converted 80% of their pages to enable search features, and has seen a 35% increase in visits."^[4]

Those gains are about rich results and click-through in Search, not a secret AI ranking boost, and that is exactly how Google frames them. For a local business the practical read is: add the structured data that describes what you genuinely are (your business, your services, your reviews), keep it matched to the visible page, and treat it as Search hygiene rather than an AI cheat code.

The AI surfaces in plain terms

Three terms come up repeatedly, so here they are without the jargon. AI Overviews are the summarised answers that appear on some results with links out to supporting pages. AI Mode is the more conversational surface for complex, multi-part questions. Both may use a "query fan-out" technique, which Google describes as issuing "multiple related searches across subtopics and data sources" to build a response.^[2] If a single search used to take you three follow-up searches, fan-out does several of them at once, which is why more pages can be pulled in. We cover this mechanic in detail in query fan-out explained.

The third term is agents. Google's guide adds a forward-looking section describing them:

"AI agents are autonomous systems that can perform tasks on behalf of people, such as booking a reservation or comparing product specifications."^[1]

For most small businesses that is a "watch this space" item today, not a to-do, but it is coming from Google's own roadmap rather than a rumour.

Measure it in Search Console, not in a third-party "AI rank" tool

Google's guide is blunt about the tools that promise to show your "AI ranking": no third-party product can see inside its ranking or AI systems, as the table below quotes directly.^[1] It points owners instead to the Generative AI performance report in Search Console, which folds AI-feature appearances into your normal Search performance data. There is also a quiet note in the companion page that the clicks arriving via AI Overviews tend to be "higher quality (meaning, users are more likely to spend more time on the site)."^[2] Treat any tool claiming an inside line to Google's AI as marketing, and check the free first-party report instead.

What this means for a local business

Here is the circulating advice set against Google's own guide, line for line.

Nearly every paid shortcut is on Google's ignore list. What moves the outcome is unglamorous and free: useful content, a crawlable and well-structured site, and current business details. Google even reassures owners that "plenty of content thrives in Google Search (including generative AI experiences) without any overt SEO at all."^[1]

A short self-check you can run today

You do not need a tool to sanity-check the basics. Open your own site and ask: can every important page be reached by a crawler, is the main content there as real text rather than locked inside a script, does each page say something a summary could not, and do your business details (hours, services, location) match everywhere they appear? If you want the platform's own version of these questions, Google's helpful-content page asks, among others, whether the content is "written or reviewed by an expert or enthusiast who demonstrably knows the topic well".^[3] Those are the levers Google names. Everything else is noise.

Where a website check fits

The one thing an owner cannot easily eyeball is whether a machine can actually read the site cleanly: whether pages are crawlable, whether the main content is in text a parser can reach, whether the structured data matches the visible page, and whether the structure is clear enough to be understood. A free QBiz Leads AI visibility check scans your own website against exactly those machine-readiness signals and returns a plain pass or fail on each, so you can fix the foundations Google's guide says actually matter.

The check reads your website, not the engines. It does not query ChatGPT, Gemini, Perplexity, or AI Overviews, and it does not tell you whether or how any AI names your business. It tells you whether your own pages are built so that Google's systems, and other machines, can find, read, and understand them. Your own pages are what you control, and they are what the guide keeps coming back to.

Get your AI Visibility audit →

Frequently asked questions

Is there a separate "AI SEO" I need to do for Google?

No. Google's own guide states that its generative AI features "are rooted in our core Search ranking and quality systems", and that "optimising for generative AI search is optimising for the search experience, and thus still SEO". The same foundations that earn ordinary Search visibility are what make a page eligible for AI Overviews and AI Mode.

Do I need an llms.txt file to appear in Google's AI features?

No. Google's guide says you "don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them." Creating one will neither help nor harm your Google ranking.

Is structured data required to show up in AI Overviews or AI Mode?

No. Google states that "structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." It still recommends keeping structured data because it helps a page qualify for rich results in ordinary Search, where Google's own case studies report higher click-through rates.

Can a third-party tool tell me my "AI ranking" on Google?

No. Google's guide is explicit that "no third-party tool has access to our internal ranking or AI systems." Google points owners to the Generative AI performance report in Search Console, which folds AI-feature appearances into your normal first-party Search data.

What single thing helps most with Google's AI features?

Unique, useful content. Google's guide says creating content people "find unique, compelling, and useful will likely influence your website's presence in generative AI search in the long run more than any of the other suggestions in this guide." For a local business, writing from first-hand experience of the jobs you actually do carries furthest.

My pages are indexed but I never appear in AI Overviews. What am I missing?

Possibly nothing. Google's guide describes what makes a page eligible to appear, not what guarantees it. Being indexed and eligible for a snippet is the floor, not a queue you rise up. Whether an AI feature shows at all for a given search stays with Google, and there is no setting to switch on.

I already have an llms.txt file. Should I delete it?

There is no need to remove it, and no benefit in keeping it for Google. Check whether it has been treated as a substitute for pages a crawler can read. A file listing your services does nothing if the service pages themselves are thin, blocked, or missing.

How can I tell whether my own page counts as commodity content?

Read it back and ask whether someone could get the same thing from a summary of three competitors. If they could, it is commodity. What usually separates the two is whether the page contains something only you would know. That might be what a job actually costs to put right, or the fault you keep seeing on one particular model.

I already paid for AEO work. Was any of it wasted?

Split the invoice in two. Anything that made pages crawlable, clearer, better written or more current still counts, because those are the foundations the guide points at. Anything that only produced a special AI file, chunked copy, or extra markup added for AI does nothing for Google Search.

What should I ask someone selling me an AI visibility package?

Ask which part of Google's published guide their deliverable is based on, and ask for the link. If the answer is a special file, chunked pages or markup added for AI, the documentation already says those are skipped for Google Search. If the answer is content, crawlability and current business details, that is what site owners were already being told to do, under a new label.