What is GEO and how does it differ from traditional SEO?

Generative Engine Optimization (GEO) refers to optimizing content to be cited and referenced by AI-powered answer systems like Google AI Overviews, ChatGPT, and Perplexity. Traditional SEO targets ranking positions in search result pages. GEO targets inclusion in AI-generated answers. The signals are overlapping but not identical: GEO rewards verifiable statistics, expert quotes, clear E-E-A-T signals, and structured data more heavily than keyword placement.

Should I block AI crawlers from my WordPress site?

It depends on your goals. Blocking AI training crawlers (like GPTBot and CCBot) prevents your content from being used to train AI models without compensation. However, blocking indexing crawlers (like Google-Extended used for AI Overviews, or PerplexityBot) means your content will not appear in AI-generated answers. Most sites benefit from allowing indexing crawlers while potentially blocking training crawlers. Review Cloudflare AI Bot Management and robots.txt settings carefully.

What is llms.txt and does my WordPress site need one?

llms.txt is an emerging convention (proposed by Jeremy Howard) where sites provide a Markdown-formatted document at /llms.txt describing their content, key pages, and preferred AI access policies. AI systems that check for it can use it to better understand and represent your site. Only about 7–10% of domains have one as of 2026. Rank Math's AI module can generate a basic llms.txt automatically. Worth adding, but not yet a critical ranking signal.

How much has CTR dropped due to AI Overviews?

Research from 2025 shows CTR drops of 34–61% for queries where Google displays AI Overviews, compared to equivalent queries without AI answers. The effect is strongest for informational queries – exactly the type most content sites target. Zero-click searches now account for approximately 60% of all Google searches. This represents a structural shift in how organic search traffic works, not a temporary fluctuation.

Know-how

WordPress SEO in the AI Era – GEO, AI Overviews, and E-E-A-T

Published: April 15, 2026 · Author: Marcin Szewczyk-Wilgan

The rules of SEO changed more in 2024–2025 than in the previous five years combined. Google now displays AI-generated answers on approximately 50% of all searches. Studies measuring CTR on queries with AI Overviews show drops of 34–61% compared to queries without them. Zero-click searches account for roughly 60% of all Google searches. If you are running a content-driven WordPress site and your traffic strategy was built entirely on traditional SEO, you are operating on assumptions that are no longer reliable. This article examines what the AI-driven search landscape means for WordPress sites and what you can actually do about it.

The New Reality: AI Overviews and Zero-Click Search

The numbers are stark enough that understanding them precisely matters before deciding on a response strategy.

AI Overviews coverageGoogle AI Overviews (formerly Search Generative Experience) now appear on approximately 50% of searches in the US, with lower but growing prevalence in European markets. They dominate informational queries: how-to questions, definitions, comparisons, and explanations – the query types that most content sites built their traffic on.

CTR impactA 2025 study found CTR drops of 34–61% for queries with AI Overviews present. The effect is not uniform: commercial and navigational queries (branded searches, product searches with buying intent) are less affected. Purely informational queries – the long tail that powers most content sites – are most heavily impacted.

Zero-click searchesApproximately 60% of all Google searches in 2025 ended without a click to any website. This includes AI Overviews, featured snippets, Knowledge Panel answers, and other SERP features that answer the query directly. The trend has been growing for years; AI Overviews accelerated it significantly.

AI citation patternsResearch from Princeton, Georgia Tech, and the Allen Institute found that AI answer systems cite 5–16 sources per response. Reddit appears in over 20% of AI-cited domains – a consequence of AI systems' preference for first-person experience and community consensus. Sites with clear E-E-A-T signals are cited more frequently than generic content farms.

Generative Engine Optimization (GEO)

GEO is the emerging practice of optimizing content to appear in AI-generated answers. The same Princeton/Georgia Tech/Allen Institute research that mapped AI citation patterns also identified which content signals improve citation frequency:

Statistics (+35–40%)

Specific, verifiable data

Pages citing specific statistics, percentages, and numbers from identifiable sources appear in AI answers 35–40% more frequently than comparable content without them. AI systems prefer content that provides verifiable factual anchors. Link statistics to their source. Include the year and sample size where relevant.

Expert quotes (+40%)

Named, credentialed sources

Content including direct quotes from named experts with identifiable credentials appears ~40% more frequently in AI answers. The expert's name, affiliation, and the quote's context all matter. AI systems use expert attribution as a credibility signal. Anonymous “experts say” formulations do not provide this signal.

Keyword stuffing (−10%)

Clarity over keyword density

Content optimized for keyword density at the expense of clarity appears 10% less frequently in AI answers. AI systems evaluate semantic coherence and directness of answers, not keyword repetition. Write for the reader; do not pad content with keyword variants.

Content freshness

Age of content matters

Content older than 90 days is cited approximately 3× less frequently than equivalent recent content. For time-sensitive topics, regular content updates with explicit publication and last-modified dates are critical. Schema.org datePublished and dateModified make this visible to AI crawlers.

E-E-A-T in the AI Era

Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) was important before AI Overviews; it is now central. AI systems use E-E-A-T signals to decide which sources to cite and trust.

ExperienceFirst-hand experience with the topic being discussed. Original photographs, screenshots, personal case studies, and "I tested this and here's what happened" content signals real experience that AI-generated content cannot replicate. This is now a deliberate quality differentiator. Document your process and include evidence of direct involvement.

ExpertiseDemonstrated knowledge depth on the subject. On WordPress, this includes: author bios with credentials and professional history, links to the author's professional profiles and other published work, content that goes beyond surface-level explanations to address edge cases and nuances that only practitioners would know.

AuthoritativenessRecognition from other credible sources. Backlinks from authoritative domains, mentions in industry publications, and citations in research are the traditional signals. For AI systems, being cited by other pages that AI has already identified as authoritative creates a network effect. Building genuine relationships and industry presence matters more than link building as a pure SEO tactic.

TrustworthinessTransparency and verifiability. Named authors with real credentials. Explicit privacy policy and terms. Contact information. Date-stamped content with clear last-review dates. No affiliate disclosure omissions. HTTPS with a valid certificate. These signals tell both Google and AI systems that the site is operated by real, accountable people.

Schema.org Structured Data for AI Visibility

Structured data markup helps AI systems understand and represent your content accurately. The most relevant Schema.org types for AI visibility in 2026:

Article / NewsArticleProvides explicit metadata about content: headline, datePublished, dateModified, author, description, and image. AI systems use this to determine content freshness, attribute authorship, and understand the scope of coverage. Always include author with Person schema pointing to a real person, not just "Organization."

FAQPageMarks up question-and-answer content explicitly. AI systems use FAQ schema to extract direct answers to user questions. This is one of the highest-impact structured data types for AI answer inclusion. Structure your FAQ answers as complete, standalone responses that are accurate without surrounding context.

HowToMarks up step-by-step instructional content with explicit steps, time estimates, tools, and supplies. AI systems extract and present HowTo steps directly. For WordPress tutorials, how-to guides, and procedural content, HowTo schema is the structured data type most likely to result in AI answer inclusion.

Organization / PersonEstablishes the identity and credentials of the publisher or author. Include: name, URL, logo, social profiles, and for Person schema: job title, organization affiliation, and professional credentials. This provides the entity-level E-E-A-T signals that AI systems use to evaluate source credibility independently of page-level signals.

llms.txt and AI Crawler Management

Two practical technical measures for the AI era: llms.txt for AI discoverability and robots.txt for AI crawler control.

llms.txt

An AI sitemap for your site

Proposed by Jeremy Howard (fast.ai), llms.txt is a Markdown document at /llms.txt that describes your site's content, key pages, and guidance for AI systems. Only ~7–10% of domains have one as of 2026, but adoption is growing. Rank Math generates a basic llms.txt automatically. Worth adding as a signal, though its direct ranking effect is not yet proven.

AI indexing crawlers

Allow for AI answer inclusion

Crawlers used to index content for AI answers (Google-Extended, PerplexityBot) should generally be allowed if you want your content cited. Blocking them means your content will not appear in those AI systems' answers. These are distinct from training crawlers (GPTBot, CCBot) that use content for model training.

AI training crawlers

Your choice to allow or block

GPTBot (OpenAI), CCBot (Common Crawl), and similar training crawlers use content to train AI models. You can block them in robots.txt: User-agent: GPTBot Disallow: /. This prevents your content from being used as training data but does not affect your visibility in AI answer systems like ChatGPT search (which uses a separate indexing crawler).

Cloudflare AI Bot Management

Default blocking is significant

Cloudflare's AI Bot Management feature, enabled by default on new Cloudflare-proxied sites, blocks many AI crawlers. If you want your content indexed by AI systems, verify which crawlers Cloudflare is blocking and selectively allow the ones relevant to AI answer systems. The default blocking configuration is conservative.

Content Strategy Adjustments for 2026

The practical content strategy response to AI-driven search is less about technical tricks and more about content quality and type:

Prioritize commercial intentInformational queries ("what is X," "how to do Y") are most heavily impacted by AI Overviews. Commercial queries ("best X for Y," "X vs Y for [specific use case]," product comparisons with real test data) are less affected because they require recency, specificity, and purchase context that AI synthesis struggles to replicate reliably.

Build audience ownershipEmail lists, direct browser notification subscribers, and community channels (Slack, Discord) are traffic sources that bypass search entirely. Sites that invested in audience ownership before 2024 are far less exposed to AI-driven CTR declines. Start building these now; the window to do it before organic traffic further erodes is shrinking.

Depth over coverageOne genuinely expert article with original research, real examples, and verifiable claims outperforms ten thin AI-assisted summaries in both AI citation probability and reader trust. The scaling strategy for content volume that worked from 2015–2023 is now actively counterproductive.

Server-side renderingContent rendered client-side via JavaScript (headless WordPress, React-heavy themes) may not be indexed by AI crawlers that do not execute JavaScript. Ensure your content-critical markup is in the server-rendered HTML, not dependent on JavaScript execution. Check with curl -s https://yoursite.com | grep "your content" to verify what crawlers see without JS.

Summary

The AI transformation of search is not a temporary disruption that will stabilize back to familiar patterns. The 60% zero-click rate and 34–61% CTR drops on AI Overview queries represent a structural change to the information economy that search-dependent WordPress sites are operating in. The sites that will do well in this environment share common characteristics: real authors with real credentials, original research and data, structured markup that makes content machine-readable, technical hygiene (server-side rendering, Core Web Vitals), and audience channels that do not depend on organic search. None of these are new ideas; they are just more important now than they have ever been.