142 AI SEO Statistics October 2026

AI SEO statistics for 2026 describe two different problems. One is visibility inside AI answers, where a page can be cited without being clicked or even named. The other is the organic click, which is shrinking on the searches that now show an AI answer.

Every number on this page links to the organisation that collected it. Most of the data comes from SEO software vendors that run large studies on their own tracking databases, plus Google, Microsoft, Pew Research Center and academic work. Vendor research is named as vendor research, and where two studies measure the same thing differently both are shown. For traffic volumes from chatbots see our AI chatbot traffic statistics, for usage and zero-click rates see AI search statistics, and for the wider picture of rankings and budgets see SEO statistics.

How often do AI answers appear in search, and for which searches?

AI answers cluster around questions, problems and long queries, they swing with each model change, and AI Mode builds its answers differently from AI Overviews.

Which kinds of searches trigger an AI Overview?

Problem-solving and specific-question searches, long queries and informational searches do. Broad topic research and brand searches mostly do not, though that gap is closing.

• AI Overviews appeared for 29.9% of 10,000 US desktop keywords but only 11.5% of total search volume (Authoritas, Jan 2025).
• Problem-solving searches showed one 74% of the time and specific-question searches 69%, while broad topic research showed one for 3.37% (Authoritas, Jan 2025).
• Keywords with 501 to 2,400 monthly searches triggered one for 42% of terms, against 36% for terms with 101 to 500 searches (Authoritas, Jan 2025).
• Telecoms had the highest sector share at 56% and Beauty and Cosmetics the lowest at 14% (Authoritas, Jan 2025).
• 33.3% of non-brand keywords showed an AI Overview, against 19.6% of brand terms and 69.8% of informational brand terms (Authoritas, Jan 2025).
• 36% of searches that contained both a noun and a verb produced an AI summary (Pew Research Center, Jul 2025).
• The median AI summary ran 67 words, and 6% of the sources linked in summaries were .gov sites, against 2% in standard results (Pew Research Center, Jul 2025).

How much did AI Overview coverage change over the past year?

It rose for most of a year, then dipped, and the size depends on which keyword set a vendor tracks. Treat every rate as a snapshot.

• BrightEdge's tracked queries went from 26.6% to 44.4% with an AI Overview between May 2024 and September 2025 (BrightEdge, Sep 2025).
• Education coverage grew from 15.4% to 85.2% over that period, while eCommerce fell from 27.2% to 18.5% and Finance sat at 17.9% (BrightEdge, Sep 2025).
• SE Ranking's tracked keywords went from about 28% in May 2025 to 59.75% in December 2025 and 60.85% in early January 2026 (SE Ranking, Feb 2026).
• After Gemini 3 became the default model the rate fell to 55.21%, and every one of the 20 niches saw a decrease (SE Ranking, Feb 2026).
• Just over 20% of German keywords show an AI Overview, rising to about 30% in the long tail, and 78.6% of them sit above the organic results (SISTRIX, Feb 2026).
• Among 963 French domains, the median AI Overview rate was 5.4% of queries, 29% of domains were above 10%, and health reached 22.2% (Ahrefs, Sep 2026).

How do AI Mode answers differ from AI Overviews?

AI Mode links to more domains from a side panel, shares little with AI Overviews, and increasingly cites Google itself.

• 92% of AI Mode answers carried a sidebar of links averaging about 7 unique domains, against 3 for AI Overviews (Semrush, Jul 2025).
• 1.7% of AI Mode answers had no links at all and 13.49% included a local pack (Semrush, Jul 2025).
• Commercial queries produced AI Mode responses about twice as long as informational ones (Semrush, Jul 2025).
• An AI Mode answer cites 12.6 links on average and an AI Overview 13.3, and 90.8% of AI Mode links sit in a separate block rather than in the text (SE Ranking, Aug 2025).
• AI Mode and AI Overviews shared 10.7% of URLs and 16% of domains for the same keywords (SE Ranking, Aug 2025).
• For the same prompt, AI Overviews and AI Mode cite different domains 83% of the time, a Jaccard overlap of 0.17 (SISTRIX, May 2026).
• Google.com is the most cited domain in AI Mode at 17.42% of 1.32 million citations, up from 5.7% in June 2025 (SE Ranking, Mar 2026).

What do AI answers do to organic clicks?

Clicks fall, and they fall hardest for sites whose queries trigger an AI answer most often. Brand searches and position 1 do not protect a page the way they used to.

How large is the click-through loss?

Between 6% and 25% on average, depending on how exposed a site is. A few percentage points of CTR is the typical cost, and the worst-hit sites lose far more.

• Across 963 French domains, click-through rate fell 3.4% in nine days, and 82.2% of highly exposed domains lost CTR (Ahrefs, Sep 2026).
• The slope between a domain's AI Overview rate and its CTR change was -0.9, so a site with 15% of queries affected could expect to lose about 15% of CTR (Ahrefs, Sep 2026).
• The panel median was -5.7% CTR and -8.8% clicks, with almost one site in two losing more than 10% of daily clicks and one in twenty losing half (Ahrefs, Sep 2026).
• Keywords that triggered an AI Overview lost 15.49% of CTR on average across 700,000 keywords on 10 sites in five industries (Amsive, Apr 2025).
• Non-branded keywords lost 19.98% of CTR, while branded keywords that triggered an AI Overview gained 18.68%, although only 4.79% of branded keywords triggered one (Amsive, Apr 2025).
• Keywords ranking outside the top 3 lost 27.04% of CTR, and keywords showing both an AI Overview and a featured snippet lost 37.04% (Amsive, Apr 2025).
• Across all German searches, 57% of searches led to an organic click on a page without an AI Overview, against 33% with one (SISTRIX, Feb 2026).

Which sectors and sites lose the most clicks?

Health, family and lifestyle content, and reference sites. Recipe and shopping searches lose little.

• Germany lost about 265 million organic clicks a month to AI Overviews, an average click loss of 6.6% across all keywords (SISTRIX, Feb 2026).
• By category, Family and Baby lost 24.31% of clicks and Health 21.03%, while Recipes lost 1.07% (SISTRIX, Feb 2026).
• Wikipedia lost 31.6 million clicks a month, the largest absolute loss, and specialised health portals lost more than 30% proportionally (SISTRIX, Feb 2026).
• French health sites lost 20.4% of CTR, and .fr domains lost 12.3% of clicks at the median against 8.8% for the whole panel (Ahrefs, Sep 2026).
• Review platforms lost between 76.5% (Gartner Peer Insights) and 92.2% (TrustRadius) of organic traffic from early 2024 to the end of 2025 (SE Ranking, Jan 2026).
• Even so, 34.5% of AI Overviews for commercial keywords cited at least one review platform, rising to 49% for searches containing "review" and falling to 17.1% for "best" and "top" lists (SE Ranking, Jan 2026).

How do people behave when an AI answer is on the page?

They re-read it, compare it and hesitate. In AI Mode they mostly accept the shortlist and leave.

• A clickstream study of about 846,000 US Google sessions found cursors spread across 83% of the viewport with an AI Overview, against 66% without (Growth Memo, May 2026).
• Users kept the cursor still 44% of the time with an AI Overview, against 29% without (Growth Memo, May 2026).
• Among users who reversed scroll direction, 47.5% of their scrolling went back up the page with an AI Overview, against 27% without (Growth Memo, May 2026).
• At 21 seconds, only 12% of navigational searchers were still on a page with no AI Overview, but with one, all five intent types sat between 41.9% and 48.5% (Growth Memo, May 2026).
• Cursor activity on navigational brand searches rose 40% when an AI Overview was present (Growth Memo, May 2026).
• In an April 2026 study of 185 high-stakes purchases, 74% of AI Mode users picked the item ranked first (Growth Memo, May 2026).

Which pages do AI answers cite, and how stable are the lists?

Mostly a short list of large platforms, then a long rotating tail. A page can rank first on Google and still be missing from the answer, and a page can be cited this week and gone the next.

Which domains do AI answers cite most?

YouTube, Wikipedia, Reddit and Google itself, with the mix changing by engine and by month.

• Across 36 million AI Overviews and 46 million citations, YouTube accounted for about 23.3%, Wikipedia about 18.4% and Google.com about 16.4% (Surfer, Aug 2025).
• For health queries, NIH led AI Overview citations at about 39%, ahead of Healthline at about 15% and Mayo Clinic at about 14.8% (Surfer, Aug 2025).
• Across 680 million citations, Wikipedia was ChatGPT's most cited source at 7.8%, Reddit led AI Overviews at 2.2% and Perplexity at 6.6%, and .com domains took more than 80% (Profound, Jun 2025).
• Reddit was the top cited domain on Perplexity at 4%, second on ChatGPT Search at 13% and third on Google AI Mode at 9% in a study of 248,000 Reddit URLs (Semrush, Nov 2025).
• YouTube took 9.40% of AI Overview citations after the Gemini 3 rollout, ahead of Reddit, Facebook, Quora and Indeed (SE Ranking, Feb 2026).
• ChatGPT's citations of Reddit fell from about 1 in 5 answers in March 2026 to 1 in 43 by late September, and Wikipedia's from 1 in 7 to 1 in 35 (Ahrefs, Oct 2026).

Do AI answers cite pages that rank in Google's top 10?

Often, but far from always, and the answer depends on the engine.

• Perplexity's citations overlapped Google's top 10 by 91.44% of domains and 82.34% of URLs, AI Overviews by 85.79% and 67.39%, AI Mode by 53.68% and 35.41%, and ChatGPT by 44.65% and 27.73% (Semrush, Jul 2025).
• Across 9,721 keywords, AI Mode answers shared an average of 2.5 URLs with the organic top 10, and 21.9% of their domains matched (SE Ranking, Aug 2025).
• 82.1% of keywords had some URL overlap with the top 10 and 17.9% had none (SE Ranking, Aug 2025).
• The links in AI Mode's occasional organic-style block were far closer to the rankings: 87.3% came from the top 10 (SE Ranking, Aug 2025).
• Brand recommendations differed across ChatGPT, AI Overviews and AI Mode for 61.9% of queries, and only 17% returned the same brands on all three (BrightEdge, Aug 2025).

How does ChatGPT choose its sources?

It searches less often than people assume, and when it does, it favours official pages and its own search index over forums.

• Of 1.4 million ChatGPT 5.2 prompts, ChatGPT cited only about 50% of the URLs it retrieved, and 88% of cited URLs came straight from its search index (Ahrefs, Apr 2026).
• Reddit made up 67.8% of the URLs ChatGPT retrieved but did not cite, and was cited at a rate of just 1.93% (Ahrefs, Apr 2026).
• The median age of a cited page was about 500 days, and search results with natural-language slugs were cited 89.78% of the time against 81.11% for others (Ahrefs, Apr 2026).
• When ChatGPT searched, it ran about 1.6 fan-out queries per response, and 75.9% of the common site: searches pointed at a brand's own site (Ahrefs, Oct 2026).
• For factual questions in a 20,000-prompt test, 84.4% cited at least one official source, against 68.3% for comparison questions (Ahrefs, Oct 2026).
• ChatGPT cited an average of 3 to 4 domains per response against 14 to 16 in AI Mode, and 68% of its stable core sources for German queries were in English (SISTRIX, May 2026).

How stable are the sources AI answers cite?

Not very. Fixed cores exist, but the surrounding sources rotate, and a model update can remove a large share of them at once.

• About 70% of the pages ranking in AI Overviews changed over two to three months in 11,203 US keywords, and AI Overview rankings moved independently of organic rankings (Authoritas, Jan 2025).
• Only 9.2% of AI Mode URLs matched across three runs of the same query on the same day (SE Ranking, Aug 2025).
• Across 80,000 prompts per platform, 40% to 60% of cited domains differed one month later, and 70% to 90% differed when January was compared with July (Profound, Jul 2025).
• In 86.5% of AI Mode prompts a stable core of one to five domains stayed in place over 17 weeks while 89% of the other domains rotated weekly (SISTRIX, May 2026).
• In 53% of AI Overview prompts not one source changed over the 17 weeks, and only 1.4% of cited news articles stayed in the citation set (SISTRIX, May 2026).
• 46.3% of the domains cited before Gemini 3 no longer appeared in AI Overviews afterwards, and unique cited domains fell 14% (SE Ranking, Feb 2026).
• AI Overviews kept 24% of cited URLs unchanged in BrightEdge's test, against 20% for AI Mode (BrightEdge, Aug 2025).

What makes a page more likely to be cited or mentioned?

Clear answers early on the page, evidence of expertise and a recognisable brand discussed elsewhere. Markup and link counts show weak effects in the studies.

Which content traits go with AI citations?

An answer at the top, expert signals, question-and-answer structure and a page that matches the specific intent.

• Comparing 304,805 URLs cited by ChatGPT Search, AI Mode and Perplexity with 921,614 Google-ranking URLs, clear summarization scored 32.83% higher and E-E-A-T signals 30.64% higher on cited pages (Semrush, Jan 2026).
• Q&A format scored 25.45% higher, section structure 22.91% and structured data elements 21.60%, while a non-promotional tone scored 26.19% lower (Semrush, Jan 2026).
• Pages that confirmed and answered the query early were cited 45% of the time against 23% for pages that did not, in Surfer's sample (Surfer, Jun 2026).
• Early query confirmation carries a weight of 28.6% for AI Mode, 22.3% for AI Overviews, 17.9% for ChatGPT and 13.7% for Perplexity in Surfer's rubric (Surfer, Jun 2026).
• Intent queries cited intent pages 84.1%, 86.2% and 89.1% of the time, while 64% of citations for broad primary-category queries went to local pages (Yext, Jul 2026).
• Anthropic's model cited the intent page 96.9% of the time, Perplexity 93.0%, Gemini 84.2% and OpenAI 68.4% (Yext, Jul 2026).
• In German AI Mode, 85% of core URLs were evergreen against 77% of peripheral ones (SISTRIX, May 2026).

Do links, authority and technical markup matter?

Link quality has a moderate link to AI visibility. Schema, slug length and domain scores show weak or no direct effect.

• Semrush found a Pearson correlation of 0.65 between Authority Score and AI mentions across 1,000 domains, and 0.23 between Authority Score and AI share of voice (Semrush, Oct 2025).
• Follow and nofollow links correlated almost identically with AI mentions, at Spearman 0.509 and 0.504, and image links beat text links, 0.415 against 0.334 Pearson (Semrush, Oct 2025).
• Gemini and ChatGPT weighted nofollow links slightly more, while Google AI Overviews and Perplexity leaned toward follow links (Semrush, Oct 2025).
• Of 378,000 cited URLs analysed for structure, slugs of 21 to 25 characters earned the most citations, about 87,000 (Semrush, Jan 2026).
• Open Graph appeared on about 60% of pages cited by AI Mode and about 40% of those cited by ChatGPT Search, and JSON-LD on about 40% and 30% (Semrush, Jan 2026).
• 97% of top-ranking Google results now use schema markup, up from 93%, and the average results page carries 18.5 schemas, up from 9.6 (BrightEdge, Sep 2025).

Do brand mentions, reviews and press coverage matter?

Yes, and they work differently from citations. Being linked and being named are separate outcomes.

• 13.2% of domain appearances were both cited and mentioned and 25.1% were mentioned without a link, in 3,981 appearances across 115 prompts in 14 countries (Semrush, Jun 2026).
• Gemini named brands in 83.7% of appearances but linked them only 21.4% of the time, while ChatGPT linked 87% and named 20.7% (Semrush, Jun 2026).
• Comparative queries produced a 43.3% brand mention rate, against 18% for informational queries, which earned an 89.3% citation rate (Semrush, Jun 2026).
• In 1,094 categories tracked in ChatGPT, brands were cited in 74% of appearances close to their core expertise and 50% of distant ones, and named in 44% against 25% (Semrush, Aug 2026).
• Brands with no Trustpilot profile had a median AI citation rate of 1%, against 53.5% for brands with a minimal profile of 1 to 13 reviews, in a study of 804,491 responses that Trustpilot later promoted (Seer Interactive, May 2026).
• Coverage on top news sites had a correlation of about 0.07 with ChatGPT brand mentions in Seer's test on the model's earlier training data (Seer Interactive, Mar 2025).
• Controlled experiments across verticals and languages found a systematic bias in AI search toward earned media over brand-owned and social content, in contrast to Google's more balanced mix (arXiv, Chen et al., Sep 2025).
• Self-promotional pages were cited, yet 43% of answers that cited a conference page still never mentioned the conference, in a 9,886-answer experiment by a vendor testing its own content (Ahrefs, Jul 2026).

How does AI-written content perform in search, and what does Google say?

AI content ranks, but rarely at the very top, and Google's guidance has not changed: it judges quality and intent, whatever tool made the page.

Does AI-written content reach the top results?

It reaches page one, but position 1 is mostly human-written, and sites built only on AI articles tend to fade.

• AI-generated pages made up 16.9% of position 4 results, well above their share at position 1, using GPTZero to classify 42,000 blog posts (Semrush, Apr 2026).
• Position 1 results were 8 times more likely to be human-written in that dataset (Semrush, Apr 2026).
• In a 2024 analysis of 20,280 top-20 URLs, purely AI-generated pages were about 3% of results and pages with more than 50% AI content 12% (Graphite, Jun 2024).
• Pages with minimal or no AI content made up 88% of the first 20 positions, and human-written pages' best position fell in the top five half the time (Graphite, Jun 2024).
• SE Ranking published 2,000 AI-generated articles on 20 new sites, and Google indexed about 71% of the pages in the first month (SE Ranking, Mar 2026).
• Around month three the share of those pages in the top 100 fell from 28% to 3%, and most sites stayed at low visibility for 16 months (SE Ranking, Mar 2026).
• Six AI-assisted articles on SE Ranking's established blog earned about 555,000 impressions and 2,300 clicks over a year, and three ranked in the top 10 (SE Ranking, Mar 2026).

What do SEO professionals say about AI content?

Most use AI for text work, believe it ranks about as well as human writing, and keep humans in the loop.

• 72% of SEO professionals who use AI content say it ranks as well as or better than human-written content, up from 64% in 2024, in a survey of 224 SEOs (Semrush, Apr 2026).
• The share who say it ranks worse rose from 9% to 13%, and those unsure fell from 27% to 15% (Semrush, Apr 2026).
• 64% of SEOs use a human-led, AI-assisted workflow, 23% create content without AI, and 87% report content that is human-created or human-led (Semrush, Apr 2026).
• At least 65% use AI for research, editing and on-page optimization, but only 28% for visuals, 15% for translation and 9% for video or audio (Semrush, Apr 2026).
• 70% name speed as the top benefit of AI, 62% brainstorming, and only 19% say it improves content quality (Semrush, Apr 2026).
• In a survey of 371 SEO professionals, 58% were Hybrid Strategists using AI to support human-written content, 49% Authority Builders and 22% AI-Heavy Adopters (Search Engine Journal, Sep 2025).

What does Google say about AI content and AI features?

That AI content is judged like any other content, that scale without value is spam, and that there is no special markup or file for AI features.

• Google says using generative AI to produce many pages without adding value for users may violate its scaled content abuse spam policy (Google Search Central, Dec 2025).
• Google advises manually fact-checking and reviewing all AI-generated content before publishing because generative models can produce inaccuracies (Google Search Central, Dec 2025).
• To appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show with a snippet, with no additional requirements (Google Search Central, Oct 2026).
• Google says AI Overviews are shown only when they add to classic Search and often do not trigger (Google Search Central, Oct 2026).
• Google says there is no ideal page length, no need to chunk content for AI, and that seeking inauthentic mentions is less helpful than it seems (Google Search Central, Jul 2026).
• Google describes AEO and GEO as SEO from its perspective, and says structured data is not required for generative AI search (Google Search Central, Jul 2026).
• An academic benchmark accepted at KDD 2024 found that optimization methods can boost a page's visibility in generative engine responses by up to 40% (arXiv, Aggarwal et al., Jun 2024).

How are SEO teams adopting, funding and measuring AI search work?

Most teams have changed strategy but not their operating model. Tracking is largely manual, and the tools they use disagree with each other.

How many teams have changed their SEO strategy for AI?

Most have adjusted strategy, treat AI search as part of SEO, and are still working out who owns it.

• 85% of marketers say AI changed their search strategy, 32% significantly, in a survey of 481 marketers and SEO professionals (Semrush, Jun 2026).
• 77% describe AI search as part of SEO, with 55% calling it an extension of SEO, and only 15% still treat it as a separate channel (Semrush, Jun 2026).
• Ownership is scattered: a dedicated AEO or GEO specialist leads at 18%, SEO at 16%, content at 15% and shared ownership at 14% (Semrush, Jun 2026).
• 37% say competitors are mentioned more often than their brand in AI answers, 30% say their brand is described inaccurately and 29% say its positioning is unclear (Semrush, Jun 2026).
• 49% name content creation as their top investment area, 46% brand visibility across channels, 38% AEO and GEO, 25% AI visibility tools and 36% traditional SEO (Semrush, Jun 2026).
• US searches for "generative engine optimization" were up 84% year over year to 7,200 a month, and "ai visibility tools" up 266% (Ahrefs, Sep 2026).

How do SEO teams use AI for keyword research, content and analysis?

For text, analysis and clustering first, then images. Use of AI for analysis and clustering roughly doubled in a year.

• 29% of marketers using AI now ask AI to check the quality of AI-generated content, up from 14% a year earlier (Ahrefs, Sep 2026).
• AI use for keyword clustering rose from 23% to 45% and for data analysis from 28% to 60% in a year (Ahrefs, Sep 2026).
• Among AI users, 83% use Claude, 76% ChatGPT and 58% Gemini, and 68% use skill files (Ahrefs, Sep 2026).
• AI writing assistants ranked fourth among SEO tools at 42.3%, level with technical SEO tools, behind analytics and reporting tools at 56% (Search Engine Journal, Sep 2025).
• 49% of SEO professionals planned to invest in E-E-A-T and 33% in topical authority and content architecture (Search Engine Journal, Sep 2025).
• 56% of AI-using marketers say human-written content is better quality than their AI content (Ahrefs, Sep 2026).

How do teams track visibility in AI answers?

By hand, with an SEO platform or with a specialist tool, and often without a clear business metric.

• 40% of marketers rely mainly on manual prompt checks in AI tools, 38% use a traditional SEO platform and 36% a specialized AEO or GEO tool, while 13% have no consistent approach (Semrush, Jun 2026).
• Only 32% say AI search visibility is clearly measurable in terms of business outcomes, 39% say partially and 17% very difficult (Semrush, Jun 2026).
• The top friction points are limited AI visibility tracking at 34%, fragmented data at 33% and difficulty connecting data to outcomes at 32% (Semrush, Jun 2026).
• Scraped consumer interfaces triggered a web search on 88% of ChatGPT prompts and 100% of the other four products, while ChatGPT's API searched on 83% (Surfer, Oct 2026).
• A ChatGPT answer in the app cited 12.1 sources against 3.1 through the API, in 13,779 answers across 14 model and method combinations (Surfer, Oct 2026).
• The median prompt in Seer's studies reached 20 words by June 2026, up 25% in a year, and keyword-style prompts fell from 18% to 3% (Seer Interactive, Jun 2026).
• In a Seer client study, 44% of real prompts included brand names (Seer Interactive, Feb 2026).

What do Google and Bing report about AI answers?

Google reports AI impressions and Bing reports AI citations, and neither report covers clicks from AI answers on their own.

• Google launched Search Generative AI performance reports in Search Console on 3 June 2026, showing impressions, pages, countries, devices and dates for AI Overviews, AI Mode and Discover (Google Search Central, Jun 2026).
• Google rolled these reports out to all websites worldwide by 31 August 2026 (Google Search Central, Jun 2026).
• Microsoft's AI Performance report in Bing Webmaster Tools, in public preview since 10 February 2026, shows total citations, average cited pages, grounding queries and page-level citation activity across Copilot and Bing summaries (Microsoft, Feb 2026).
• Semrush's analysis of Bing's grounding query data found that 39% of the queries used to ground AI responses were conversational questions or longer phrases (Semrush, Feb 2026).
• The Bing report measures citations only, not clicks, rankings or traffic from AI answers (Semrush, Feb 2026).

Methodology and Sources

We included a number only after tracing it to the organisation that produced it and opening that page to confirm the figure and its date. Vendor studies from Ahrefs, Semrush, SISTRIX, SE Ranking, Surfer, Seer Interactive, BrightEdge, Authoritas, Amsive, Profound, Yext and Growth Memo are included where the vendor collected the data, and are named in every citation; their keyword sets, prompt sets and country coverage differ, so rates for the same metric, such as AI Overview coverage, are shown side by side instead of averaged. Google and Microsoft statements are quoted from their own documentation. Academic work comes from arXiv. Where a study dates its data window differently from its publication, the citation uses the month of publication, or the end of the data window when the page carries no date. The page is refreshed monthly, and fast-moving figures such as AI Overview coverage and citation drift are the first to be replaced.

Sources

• Ahrefs, AI Overviews Cut CTR by 23.1% in France
• Ahrefs, Why ChatGPT Cites One Page Over Another
• Ahrefs, How ChatGPT Does Query Fan-Out
• Ahrefs, Self-promotional content experiment
• Ahrefs, 69% of Marketers Are Publishing More AI Content Than Last Year
• Ahrefs, 8 AI Marketing Trends I've Seen Firsthand in 2026
• Amsive, Google AI Overviews: New Research Reveals How to Navigate Click Drop-Off
• arXiv, Aggarwal et al., GEO: Generative Engine Optimization
• arXiv, Chen et al., Generative Engine Optimization: How to Dominate AI Search
• Authoritas, AI Overview User Intent Research
• Authoritas, SERP, Organic and AI Overview Volatility Research
• BrightEdge, AI Overview Rollout Reveals a Clear Intent Hierarchy
• BrightEdge, ChatGPT vs Google AI: 62% Brand Recommendation Disagreement
• BrightEdge, Volatility Patterns in AI Search: AIO vs AI Mode
• BrightEdge, How SEO Changed in the AI Era
• Google Search Central, AI features and your website
• Google Search Central, Guidance on using generative AI content
• Google Search Central, Optimizing your website for generative AI features
• Google Search Central Blog, Search Generative AI performance reports in Search Console
• Graphite, AI Content and Search
• Growth Memo, Users Behave Differently in AI Overviews
• Microsoft, Introducing AI Performance in Bing Webmaster Tools
• Pew Research Center, Google users are less likely to click on links when an AI summary appears
• Profound, AI Platform Citation Patterns
• Profound, AI Search Volatility
• Search Engine Journal, The State of SEO 2026
• Seer Interactive, Gemini Citations Decreased 23pp
• Seer Interactive, Study of 800K AI Responses: How Review Profiles Shape Brand Presence
• Seer Interactive, Does Being Mentioned on Top News Sites Impact AI Answer Mentions
• Seer Interactive, Are You Picking the Prompts for AI Tracking Without Observing Real Customers
• Seer Interactive, AI Visibility Is a Vanity Metric
• SE Ranking, AI Mode Research
• SE Ranking, Gemini 3 and AI Overviews
• SE Ranking, Is Google Stealing Your Clicks in AI Mode
• SE Ranking, Review Platforms in AI Overviews
• SE Ranking, AI Content Experiment
• Semrush, The Ghost Citations Study
• Semrush, Does AI Content Rank in Search
• Semrush, AI Mode Comparison Study
• Semrush, Reddit AI Search Visibility Study
• Semrush, Content Optimization for AI Search Study
• Semrush, Backlinks and AI Search Study
• Semrush, Technical SEO Impact on AI Search Study
• Semrush, ChatGPT Topical Focus Study
• Semrush, The Operational Gap in AI Search
• Semrush, Bing AI Performance Report
• SISTRIX, AI Overviews in Germany
• SISTRIX, AI Citation Drift
• Surfer, AI Citation Report 2025
• Surfer, Why Almost 40% of AI Citations Come from Your First 100 Words
• Surfer, Why You Cannot Track AI Visibility Through an API
• Yext, AI Page Citations: Specific Intents Need Specific Pages

Glossary

• AI Overview (AIO): the AI-generated summary Google shows above the regular results for some searches.
• AI Mode: Google's conversational search mode, where the results page is an AI answer with links.
• AEO: answer engine optimization, making content easy for AI assistants to find, quote and cite.
• GEO: generative engine optimization, another name for AEO focused on generative AI answers.
• Citation: a link to a page shown as a source in an AI answer.
• Mention: a brand named in the text of an AI answer, with or without a link.
• Ghost citation: a citation where the AI links the page but never names the brand.
• Fan-out query: one of several related searches an AI system runs to build a single answer.
• CTR: click-through rate, the share of people who see a result and click it.
• E-E-A-T: experience, expertise, authoritativeness and trustworthiness, Google's quality framework.
• Grounding query: the phrase an AI system uses to retrieve a page it later cites.
• llms.txt: a proposed file that lists a site's key pages for AI tools to read.