AI SEO in 2026: What Actually Changed (and What Didn't)

Google's own 2026 documentation is blunt about this: there is no special markup, no AI text file, and no content "chunking" trick required to appear in AI Overviews or AI Mode. What actually changed this year is the machinery around search, not the fundamentals of ranking — how Google selects sources for an AI answer, how it now lets publishers flag themselves as a "preferred source," how it reports AI visibility in Search Console, and how often people click through at all once an AI summary is on the page. Here's what's real, straight from Google's own pages, and what marketers are still wasting budget chasing.

What Hasn't Changed, According to Google Itself

Google's official generative AI optimization guide, last updated July 2026, opens with a line that undercuts most of the "GEO" advice circulating online: "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." In plain terms, AI Overviews and AI Mode do not run on a separate ranking system you can reverse-engineer with new tricks — they pull from the same index, built on the same crawlability, quality, and helpfulness signals Google has described for years.

The same document is explicit about several specific tactics marketers keep asking about. On special files: "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." On llms.txt by name: creating one "will neither harm nor help your site's visibility" for Google Search. On schema markup: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." And on the popular advice to break content into small, isolated chunks for AI parsing: "There's no requirement to break your content into tiny pieces for AI to better understand it." Google's AI features page adds the plainest possible summary: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."

Separately, Google Search Advocate John Mueller addressed llms.txt directly in June 2026, calling its usefulness "purely speculative for now" and noting that "the file has existed for years, yet none of the AI systems use it." His practical suggestion wasn't to add the file defensively — it was to only bother if a specific AI platform that actually sends you traffic asks for one.

What Actually Changed in 2026

None of that means nothing changed. Four things genuinely shifted this year, and they matter more than any markup trick would have anyway.

1. Google explained "query fan-out." Google's documentation now describes how AI Overviews and AI Mode work under the hood: instead of running one search, the systems issue "multiple related searches across subtopics and data sources" and assemble the answer from a wider, more diverse set of pages than a classic results page would show. For a marketer, the practical implication is that a page can get pulled into an AI answer for a related sub-question it never explicitly targeted — which rewards genuinely comprehensive pages over ones narrowly optimized for one exact keyword phrase.

2. "Preferred Sources" launched, and it touches AI surfaces. Google's Preferred Sources feature (most recently updated in Search Central's documentation in August 2026) lets a signed-in user mark a domain as a preferred publication. Marked sites can then show a "preferred" badge — and per Google's own documentation, that badge can appear "in AI Mode and AI Overviews," not just in Top Stories. Eligibility is domain- or subdomain-level only; a subdirectory like a single blog section isn't eligible on its own. This is new leverage for established publishers with a returning audience, not something a brand-new blog can manufacture overnight.

3. Search Console got an AI-specific report. In June 2026, Google Search Central announced generative AI performance reports inside Search Console, giving site owners visibility into how their pages perform specifically in AI Overviews and AI Mode, separate from classic Search. Before this, marketers were mostly guessing at AI visibility from indirect traffic patterns.

4. Clicks behave differently when an AI summary appears. This one predates 2026 but keeps compounding: a Pew Research Center study of U.S. adults' actual browsing behavior (tracked in March 2025, published July 2025) found that people who encountered a Google AI summary clicked through to a traditional search result in just 8% of visits, versus nearly 15% of visits when no AI summary appeared — and only 1% of visits involved clicking a link inside the AI summary itself. Google's own AI Overviews documentation frames this differently, saying the feature is "only shown when our systems determine that it is additive to classic Search," but the Pew data is the clearest independent evidence available on what happens to click behavior once it does appear.

The AI-SEO Myth Grid

A simple way to check any "GEO hack" you're being sold: does Google's own documentation actually say this helps?

One honest caveat: this grid reflects Google's own documented position for Google Search specifically. Other AI answer tools — ChatGPT search, Perplexity, Copilot — are run by different companies with different, less publicly documented crawling and citation systems, so a tactic that does nothing for Google isn't automatically proven useless everywhere else. Mueller's own advice reflects that nuance: build a file like llms.txt only if a specific platform that sends you real traffic asks for it, not as a blanket defensive move.

What This Actually Means You Should Do

• Confirm indexability first. If Google can't crawl and index a page normally, it can't appear in a classic result or an AI Overview — check Search Console's Coverage report before anything else.
• Check the new Gen AI performance report in Search Console. It's the closest thing to real AI-visibility data you have, and it's free.
• Check your Preferred Sources eligibility if you're a domain-level publisher with a returning audience. It's a real, documented lever — but it only applies at the domain or subdomain level, so a single blog post can't use it on its own.
• Write for the fan-out, not just the exact keyword. Since Google now explicitly issues multiple related sub-queries to build an AI answer, cover the adjacent questions a reader would naturally ask next, in the same article.
• Assume fewer clicks even when you rank. With Pew's data showing an AI summary roughly halves click-through, treat owned channels — an email list in particular — as the backup plan for reaching people who see your brand in an AI answer but never click. See our guide to building an email list from zero for the exact mechanics.
• Stop paying for llms.txt "optimization" as a standalone service. Per Google's own documentation and Mueller's public comments, it does nothing for Google visibility today. Spend that budget on the content and crawlability basics above instead.
• If you want cross-platform visibility tracking, use a tool built for it, not a file. Rank trackers like Surfer have expanded to monitor how pages surface across Google, Gemini, ChatGPT and Perplexity — see our roundup of AI tools every marketer should be using for what that actually costs before you buy in.

Worked Example: Auditing a Small Marketing Blog

Jordan runs content for a 12-person marketing agency and was recently pitched a "$1,200 GEO audit" that promised to add llms.txt, restructure every blog post into short Q&A chunks, and inject FAQ schema on every page to "get cited by AI." Running the claims through the myth grid above: the llms.txt add-on does nothing for Google per Google's own documentation; the chunking rewrite isn't required per the same source; and while FAQ-style schema isn't harmful, Google is explicit that no special schema is needed to appear in AI features. Running that $1,200 through the same lens as our AI tool budget framework made the decision easy: money spent on tactics Google's own documentation says do nothing isn't a discount, it's a full loss. Jordan declined the audit and instead spent an afternoon in Search Console confirming every important page was indexed, checked the new Gen AI performance report to see which existing pages already showed up in AI Overviews, and added one genuinely new comparison page covering a question client prospects kept asking that the agency's site never directly answered. That's the entire list of things Google's own documentation says actually matters.

Frequently Asked Questions

Does llms.txt help my site show up in ChatGPT or Perplexity, even if it doesn't help Google? Google's documentation only speaks to Google Search. Other AI platforms use different systems that aren't as publicly documented, so it's not verifiable either way from the sources here — Mueller's suggestion is to build one only if a platform that actually sends you traffic specifically asks for it.

Do I need FAQ schema or How-To schema for AI Overviews? No. Google's generative AI optimization guide states structured data isn't required and there's no special schema.org markup needed for generative AI search specifically. Structured data can still help with other classic rich-result features, which is a separate topic from AI Overviews.

Is being featured in an AI Overview guaranteed to hurt my traffic? Not guaranteed, but the direction is well-documented: Pew Research's 2025 browsing-data study found meaningfully lower click-through when an AI summary appeared (8% vs. 15% of visits). Google frames AI Overviews as only appearing when "additive" to Search, but that's a decision about answer quality, not a promise about your click-through rate.

What's the single highest-leverage thing to check first? Indexability. None of the AI-era changes matter if the page isn't crawled and indexed in classic Search to begin with — that's still the floor everything else is built on, per Google's own guidance.

Sources: Google, "Google's Guide to Optimizing for Generative AI Features on Google Search" — Search Central Documentation (updated July 2026); Google, "AI Features and Your Website" — Search Central Documentation; Google, "Guide to Preferred Sources in Google Search for Web Publishers" — Search Central Documentation (updated August 2026); Google, "Introducing Search Generative AI performance reports in Search Console" — Search Central Blog (June 2026); Pew Research Center, "Do people click on links in Google AI summaries?" (July 2025, browsing data from March 2025); Search Engine Journal, reporting John Mueller's June 2026 statement on llms.txt. Last reviewed: August 2026.