Optimizing a WordPress Site for AI Search and AI Overviews in 2026

There is more confidently stated nonsense about optimizing for AI search than about almost any other topic in SEO right now. Add an llms.txt file. Chunk your content into tiny sections. Add special schema. Block Google-Extended to opt out of AI Overviews. Rewrite everything in question-and-answer format so the models can parse it.

Nearly all of that is either unsupported or actively wrong, and in several cases Google's own documentation says so in plain language. Meanwhile the things that genuinely matter, being crawlable, being indexed, being snippet-eligible, and understanding which controls actually govern which system, get much less attention because they're unexciting.

This is what the official documentation from Google, OpenAI, Anthropic, Apple, and Cloudflare actually says, what credible research shows about the traffic impact, and what a WordPress site owner should practically do.

TL;DR

• Google's documentation states directly: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." Eligibility requires that a page is indexed and eligible to be shown with a snippet.
• Blocking Google-Extended does not remove you from AI Overviews. Per Google's own crawler documentation, Google-Extended governs Gemini and Vertex AI training and grounding, and explicitly "does not impact a site's inclusion in Google Search nor is it used as a ranking signal." The documented control for AI Overviews is nosnippet or max-snippet.
• No major AI provider has committed to reading llms.txt. Google's John Mueller was asked in January 2026 whether Google's own use of the files constituted an endorsement and answered, "to be direct, no."
• Google says there is "no special schema.org structured data that you need to add" for AI features, and no requirement to break content into tiny pieces or write differently for generative search.
• The most credible click-impact research, from Pew, found users clicked a traditional result in 8% of visits with an AI summary present versus 15% without. Treat the much larger figures circulating in SEO content with suspicion.

What Google Officially Says

Start here, because it makes most of the surrounding industry noise easier to filter.

Google's Search Central documentation on AI features and your website is unusually blunt. Its stated positions include:

• "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
• "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)."
• "There's no requirement to break your content into tiny pieces for AI to better understand it."
• "You don't need to write in a specific way just for generative AI search."
• "There's also no special schema.org structured data that you need to add."

On eligibility, Google is specific: "To be eligible to be shown in generative AI features on Google Search, a page must be indexed and eligible to be shown in Google Search with a snippet." And the content has to be reachable: "Google Search generative AI models use publicly accessible, crawlable content."

The stated reason none of this requires new tactics is architectural. Google says its generative AI features "are rooted in our core Search ranking and quality systems," which is why "the best practices for SEO continue to be relevant."

That is the whole official answer, and it's worth sitting with before spending money on anything marketed as AI SEO. The gap between what Google documents and what the AI-SEO industry sells is large.

The Controls That Actually Exist

There are real technical levers here. They're just not the ones most articles name.

nosnippet and max-snippet

These are the documented controls over whether your content feeds AI features.

nosnippet prevents any text snippet or video preview across Search, Images, and Discover, and Google's robots meta tag documentation explicitly adds that it "will also prevent the content from being used as a direct input for AI Overviews and AI Mode."

max-snippet:[number] caps snippet length, and Google states it "will also limit how much of the content may be used as a direct input for AI Overviews and AI Mode." Setting max-snippet:0 disables snippets entirely; -1 means no limit. Google's documentation notes an exception where a publisher has separately granted permission for content use.

The trade-off is unavoidable and severe: because AI Overview eligibility is tied to snippet eligibility, opting out of AI Overviews with these directives also removes your normal search snippets. You will look worse in regular search results. For most sites that's a bad trade, but it's the documented mechanism, and it's better to know that clearly than to think there's a costless opt-out.

data-nosnippet marks specific regions of a page as ineligible for snippet use, applied to span, div, or section elements. Worth a caveat: Google's documentation explicitly ties nosnippet and max-snippet to AI Overviews and AI Mode but does not make the same explicit statement about data-nosnippet. Don't assume it behaves identically for AI features.

Google-Extended: The Most Misreported Control

This one is worth getting exactly right, because the wrong version of it is everywhere.

Google's crawler documentation describes Google-Extended as "a standalone product token that web publishers can use to manage whether content Google crawls from their sites may be used for training future generations of Gemini models." It covers Gemini Apps training, Vertex AI API training, and grounding in Gemini Apps and Vertex AI.

And then, explicitly: it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search."

So blocking Google-Extended in robots.txt stops your content being used for Gemini model training and grounding. It does not remove you from AI Overviews or AI Mode in Google Search, which run off the standard Googlebot crawl and are governed by the snippet directives above.

If you've blocked Google-Extended believing you opted out of AI Overviews, you didn't. You opted out of something else entirely.

The AI Crawlers, From Their Own Documentation

Different bots, different purposes. Blocking indiscriminately has consequences worth understanding.

OpenAI documents four:

• GPTBot crawls content for training OpenAI's generative models. Disallow it to opt out of training use.
• OAI-SearchBot powers ChatGPT's search and answer surfacing. Disallowing it prevents you appearing in ChatGPT search answers, though navigational links may still appear. OpenAI notes robots.txt changes take roughly 24 hours to propagate.
• OAI-AdsBot validates landing page safety for ads submitted to ChatGPT; that content isn't used for training.
• ChatGPT-User fires on live, user-initiated actions inside ChatGPT rather than autonomous crawling. OpenAI's own documentation notes that "robots.txt rules may not apply" here, since it's triggered by a user request.

The critical distinction: blocking GPTBot opts you out of training. Blocking OAI-SearchBot opts you out of being cited in ChatGPT's answers. Many sites blocked both when they only meant the first.

Anthropic documents three: ClaudeBot collects content potentially used for training, Claude-User fires when a user asks Claude to read a specific site, and Claude-SearchBot navigates the web to improve search result quality. All are blockable individually via standard robots.txt directives, repeated per subdomain. Anthropic states it respects industry-standard robots.txt directives and supports Crawl-delay for rate limiting rather than full blocking, and notes that IP-based blocking is unreliable.

Apple operates Applebot, which powers Spotlight, Siri, and Safari search, and Applebot-Extended, which doesn't crawl independently but controls whether already-crawled data trains Apple's generative models. Apple's documentation confirms that "webpages that disallow Applebot-Extended can still be included in search results," making it the same training-only opt-out pattern as Google-Extended.

On other crawlers: Perplexity's and Microsoft's AI-specific crawler behavior isn't something I'd state with confidence here. Bing Webmaster Tools added an AI Performance report in public preview in early 2026, but the underlying crawler specifics should be checked against Microsoft's and Perplexity's own current documentation rather than any third-party summary, including this one.

llms.txt: The Thing Everyone Recommends That Nobody Reads

The llms.txt proposal came from Jeremy Howard in September 2024: a markdown file at /llms.txt intended to help language models use a site at inference time. It's a reasonable idea, and it has picked up support in various documentation tooling.

What it has not picked up is adoption by any major AI provider or search engine as a signal they actually consume.

Google has been explicit. Its optimization guidance groups llms.txt with the tactics you don't need. In January 2026, John Mueller was asked publicly whether the presence of llms.txt files on Google's own properties amounted to an endorsement. His answer: "I'm tempted to say something snarky since this has come up so often, but to be direct, no." He explained those files had appeared automatically through CMS support and were subsequently removed from Google's developer docs, and separately compared llms.txt to the deprecated keywords meta tag, noting that no AI services used it and bots weren't requesting the file.

If you've already added one, it costs nothing to leave it. If a plugin or agency is selling llms.txt generation as an AI visibility feature, that's the tell.

What the Traffic Data Actually Shows

This is where fabricated statistics are most common, so here are the two figures worth citing and the reason to distrust the rest.

Pew Research Center, in a study published July 2025, tracked browsing data from a panel of around 900 US adults across 68,879 Google searches from March 2025, of which 12,593 produced AI summaries. The finding: users clicked a traditional web result link in 8% of visits when an AI summary was present, versus 15% when it was not. Only 1% of visits with an AI summary involved clicking a link inside the summary itself.

That's a transparent methodology with a real sample, and it's the number to use.

Google's own position, stated by its VP and Head of Search in August 2025, is that "total organic click volume from Google Search to websites has been relatively stable year-over-year" and that Google is sending "slightly more quality clicks" than a year prior, defining a quality click as one where the user doesn't quickly bounce back. Google does acknowledge redistribution, "resulting in decreased traffic to some sites and increased traffic to others."

These two things are not necessarily contradictory. Per-query click rates can fall while total click volume holds steady if query volume rises, which Google says it has: at its 2026 developer conference, Google reported AI Mode surpassing 1 billion monthly users with queries "more than doubling every quarter since launch."

The various "AI Overviews cut clicks by 58%" or "61%" figures circulating in marketing content come from a mix of methodologies, vary wildly, and are disputed by Google. Don't build a business case on them.

What to Actually Do on a WordPress Site

Given all of the above, the practical work is narrower than the industry implies.

1. Make sure you're crawlable and indexed. This is the actual eligibility requirement. Check that Settings → Reading doesn't have "Discourage search engines" enabled, which also disables core's XML sitemaps. Confirm your pages are indexed in Search Console. Nothing else on this list matters if this isn't true.

2. Don't accidentally block yourself. Review your robots.txt and any security plugin's bot-blocking rules. Aggressive bot blocking configured to stop scrapers can catch OAI-SearchBot or Claude-SearchBot as collateral, quietly removing you from AI answer engines you'd probably rather appear in.

3. Decide deliberately about training versus citation. These are separate choices with separate controls. Blocking GPTBot and ClaudeBot opts out of training while leaving you eligible for citation in ChatGPT and Claude answers. Blocking OAI-SearchBot and Claude-SearchBot removes you from those answers. Pick intentionally rather than by copying a robots.txt from a blog post.

4. Write self-contained answers because it's good writing, not because AI needs it. Google says explicitly that you don't need to chunk content or write differently for generative search. But answering the main question clearly and early, using descriptive headings, and keeping claims accurate makes content better for humans and happens to make it easier to quote. The distinction matters: do it as writing craft, not as a parsing hack.

5. Use structured data for what it actually does. Google says no special schema is needed for AI features. Structured data earns rich result eligibility, which is a real benefit on its own terms. Add it for that reason.

6. Keep the site fast and stable. Crawlers on a limited budget, and AI systems fetching pages at query time, both do worse against slow or intermittently failing servers. This is boring infrastructure work rather than AI strategy, but a page that times out isn't cited by anything. Managed hosting with global edge caching handles this layer without per-site configuration.

7. If you're behind Cloudflare, know your options. Cloudflare's AI Crawl Control lets site owners monitor which AI services crawl their content and set per-crawler allow or block policies, with a Pay Per Crawl capability in closed beta that returns HTTP 402 Payment Required to crawlers without payment intent. Note that WAF and Bot Management rules take precedence, so a blocked bot stays blocked regardless of payment.

Cloudflare also published a Content Signals Policy adding a Content-Signal: line to robots.txt with three values, search, ai-input, and ai-train, letting you express preferences separately. Cloudflare is upfront that this is a signal rather than an enforcement mechanism and that some bots will ignore it, recommending WAF rules for actual enforcement.

Myths Worth Actively Rejecting

• "You need an llms.txt file." No major provider has committed to reading it, and Google has explicitly said it isn't pursuing it.
• "Block Google-Extended to opt out of AI Overviews." Google's documentation says Google-Extended doesn't affect Search inclusion. Use nosnippet or max-snippet:0 if you genuinely want out, and accept losing normal snippets too.
• "Add AI-specific schema." Google says there's no special structured data needed for AI features.
• "Break content into small chunks for AI parsing." Google explicitly says this isn't required.
• "AI Overviews cost publishers X% of traffic." The credible figure is Pew's 8% versus 15% click rate. Larger numbers vary by methodology and are contested.
• "Cloudflare's content signals legally prevent AI training." Cloudflare describes it as a non-binding signal that some bots will ignore.

One item deserves fairer treatment than outright dismissal: academic work on generative engine optimization does exist. A Princeton-led paper presented at KDD 2024 built a benchmark and found certain content-optimization strategies could boost visibility by up to 40% in generative engine responses. That's real research, but it's a controlled 2024 benchmark using specific synthetic setups, not a measurement of live AI Overviews or ChatGPT in 2026. Cite it as an interesting finding, not as a promise about your traffic.

Frequently Asked Questions

Do I need to do anything special to appear in Google's AI Overviews? Per Google's own documentation, no. It states there are no additional requirements or special optimizations. The eligibility requirement is that the page is indexed and eligible to be shown with a snippet, and that the content is publicly accessible and crawlable.

How do I opt out of AI Overviews? The documented controls are nosnippet and max-snippet. Google states both limit content being used as direct input for AI Overviews and AI Mode. The significant catch is that they also remove or truncate your normal search snippets, since AI eligibility is tied to snippet eligibility.

Does blocking Google-Extended stop my content appearing in AI Overviews? No. Google's documentation states Google-Extended governs whether crawled content trains Gemini models and grounds Gemini and Vertex AI, and that it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal." This is the most commonly misreported point in AI SEO.

Do I need an llms.txt file on my WordPress site? No provider has officially committed to reading it. Google has said it isn't pursuing llms.txt for Search, and John Mueller directly denied that Google's own use of such files was an endorsement. It's harmless to have, but it isn't an optimization.

Should I block AI crawlers like GPTBot and ClaudeBot? That depends on whether you want your content used for model training versus cited in AI answers, which are separate decisions with separate crawlers. GPTBot and ClaudeBot relate to training. OAI-SearchBot and Claude-SearchBot relate to appearing in ChatGPT and Claude search answers. Block deliberately rather than blanket-blocking.

Does schema markup help with AI search visibility? Google says no special schema.org structured data is needed for AI features. Structured data remains worth adding for rich result eligibility in regular search, which is a separate and real benefit.

Have AI Overviews reduced website traffic? The evidence is mixed and contested. Pew's study found a traditional-result click rate of 8% when an AI summary appeared versus 15% when it didn't. Google says total organic click volume has been relatively stable year over year while acknowledging traffic has shifted between sites. Be skeptical of the much larger drop figures common in marketing content.

Does writing in a Q&A format help with AI search? Google states you don't need to write in a specific way for generative search. Clear, self-contained answers and descriptive headings help human readers and make content easier to quote, which is a good enough reason on its own, but it isn't a required format.

Conclusion

The honest summary of AI search optimization for WordPress in 2026 is that it mostly isn't a separate discipline. Google says its AI features run on the same ranking and quality systems as regular search, and its documentation repeatedly rejects the idea that new files, markup, or writing styles are needed.

What is worth your attention is narrow and concrete: stay crawlable and indexed, don't let an over-eager bot-blocking rule remove you from AI answer engines by accident, understand that training opt-outs and citation opt-outs are different controls, and know that the real AI Overviews opt-out costs you your regular snippets too.

Everything else in this space is currently being sold harder than it's being evidenced. When a vendor tells you a setting unlocks AI visibility, the first question worth asking is which official documentation says so.