How AI Search Changes SEO (And What to Do About It)

Fifty-eight and a half percent of Google searches in the US now end without a single click. That number comes from SparkToro's 2024 study of tens of millions of panelists, and it means the majority of queries today are answered before anyone reaches your website. Add AI Overviews peaking at 24.61% of all queries in July 2025 (Semrush, 2025), ChatGPT reaching 700 million weekly active users by September 2025, and Perplexity processing 780 million queries in a single month — and the picture becomes clear. Search has fundamentally changed. The question is what to do about it.

This post covers what actually shifts in your SEO strategy, what the data says about content signals that AI systems favor, and the one myth (schema markup as an AI fix) that's wasting budget across the industry right now.

Key Takeaways

• AI Overviews now cover up to 25% of all Google queries; organic CTR fell 61% when present (Seer Interactive, Nov 2025)
• 58.5% of US Google searches end without a click — the open web gets only 360 clicks per 1,000 searches (SparkToro, 2024)
• Citing external sources improves AI citation visibility by up to 115%; schema markup alone has no statistically significant effect (Princeton 2024, Ahrefs 2026)
• 52% of AI Overview citations come from top-10 organic results, so SEO still determines AI eligibility

How Big Is the AI Search Shift?

Google AI Overviews appeared on just 6.49% of queries in January 2025, then surged to a peak of 24.61% in July before pulling back to 15.69% by November 2025 (Semrush, December 2025). The pullback matters: it shows Google is calibrating, not retreating. Commercial queries within AI Overviews grew from 8.15% to 18.57% during the same period, meaning AI answers are expanding into higher-intent territory.

At the same time, ChatGPT, Perplexity, and Google Gemini are absorbing search volume that would have gone to traditional Google a year ago. ChatGPT held 64.5% of the generative AI chatbot market share as of May 2026, down from 86.7% a year prior as Gemini grew from 5.7% to 21.5% (First Page Sage, May 2026). This isn't replacing Google. It's fragmenting the search landscape into multiple answer surfaces, each with its own ranking signals.

68% of marketing organizations are already changing their SEO strategy in response to AI search, with 45% pursuing multi-platform strategies covering AI Overviews, ChatGPT, and Perplexity simultaneously (BrightEdge, June 2025). The organizations that aren't moving yet are the ones who will feel this most acutely in 12 months.

What AI Search Has Done to Organic Traffic

The CTR data from Seer Interactive's November 2025 study is the clearest picture we have of the real impact. Across 3,119 queries, 42 organizations, and 25.1 million impressions, organic click-through rates fell 61% for informational queries when an AI Overview appeared. Paid CTR fell 68% on the same queries. Even queries without AI Overviews saw a 41% organic CTR decline during the same study period.

There's a split worth understanding: being cited inside an AI Overview (-49.4% organic CTR) is meaningfully better than not being cited (-65.2%). Citation doesn't restore clicks, but it reduces losses and builds brand exposure in the AI answer layer. The goal isn't to opt out of AI answers. It's to be the source those answers pull from.

Ahrefs confirms a parallel finding: position-1 results see 58% lower average CTR on queries with AI Overviews compared to equivalent queries without them, up from a 34.5% reduction measured in April 2025 (Ahrefs, December 2025). The trend is worsening, not stabilizing.

The paid search impact is the underreported story here. A 68% paid CTR decline on AI Overview queries means advertisers bidding on informational keywords are getting dramatically fewer clicks for the same spend. This is creating a real shift in how paid budgets get allocated: away from top-of-funnel informational keywords and toward mid-funnel, transactional queries where AI Overviews appear less frequently. If your business runs Google Ads on informational terms, this is as much a paid search problem as an organic one.

ChatGPT and Perplexity Are Now Search Competitors

ChatGPT reached 700 million weekly active users by September 2025 (OpenAI via DemandSage, 2025). Perplexity processed 780 million queries in May 2025 alone, growing at over 20% month-over-month (TechCrunch, June 2025). These aren't chatbots people use after failing to find something on Google. They're the first stop for a growing segment of users who prefer conversational, synthesized answers.

The citation mechanics differ across platforms. Google AI Overviews pull from indexed pages with established organic rankings. Perplexity uses real-time web retrieval and weights recency heavily. ChatGPT's web search feature prioritizes authoritative, clearly sourced content. Each platform has its own retrieval logic, but the content signals that help across all three overlap considerably: factual specificity, external citations, clear structure, and authorial credibility.

We track our own content across these platforms as part of building Frida Marketing's AI search presence. The posts that appear most consistently in Perplexity and ChatGPT results share three traits: a clear definition in the first two paragraphs, at least one named statistic with a source, and a section structure that makes each H2 extractable as a standalone answer. The posts that don't appear — even when they rank well on Google — tend to be more narrative in structure, without discrete self-contained answers that AI systems can quote.

What the Research Says: Content Signals AI Systems Favor

Princeton University's 2024 GEO paper, published at ACM KDD, is the most rigorous study available on what actually improves visibility in AI-generated search results. Tested across 10,000 queries on eight domains using a benchmark called GEO-bench, the findings are direct: citing external sources improved AI citation visibility by up to 115% for lower-ranked content, adding statistics improved visibility by 41%, and adding direct quotations improved it by 28% (Princeton/ACM, 2024).

Originality.AI's analysis of 18,767+ AI Overview-triggering keywords shows the authority pathway still runs through Google: 52% of AI Overview citations come from top-10 organic results, and the number-one ranked page has a 58% probability of being cited. The fundamental authority signals — backlinks, E-E-A-T, domain credibility — still determine who gets into the citation pool. GEO tactics improve your chances within that pool.

What this means practically: a post that quotes a named expert, cites three external studies, and includes a directly attributable statistic in every major section is structurally better positioned for AI citation than a post that covers the same ground with no external anchors. The format signals trustworthiness to the retrieval system before any human reads it.

The Schema Myth: What the Data Actually Shows

Schema markup is widely recommended as a way to improve AI Overview visibility. The recommendation sounds logical: if you're providing structured data, AI systems should be better able to extract and cite you. The reality is more complicated.

Ahrefs tracked 1,885 pages that added JSON-LD schema markup between August 2025 and March 2026, compared against 4,000 control pages. The results: Google AI Overviews showed a -4.6% change, Google AI Mode showed +2.4%, ChatGPT showed +2.2%. None of these were statistically significant (Ahrefs, 2026). Schema markup did not move the needle.

This finding cuts against a lot of agency advice being sold right now. Schema still matters for rich results in traditional SERPs, for local SEO, for e-commerce product cards, and for event listings. It is not, based on current evidence, an AI citation lever. The time you would spend implementing Article schema, FAQ schema, or Speakable schema for AI purposes is better spent on the content signals the Princeton study confirms: citations, statistics, and direct quotations. Both schema and GEO-optimized content structure take time to implement. Given limited resources, the data points clearly to which one moves the needle for AI search.

The New Playbook: What to Actually Do

The strategic shift is not to abandon SEO. It's to layer GEO on top of it, and to stop optimizing for signals that don't transfer to AI systems. Here is what that looks like in practice.

Keep doing: Everything that earns organic rankings. Top-10 organic results supply 52% of AI Overview citations. If you're not ranking, you won't be cited. Link building, technical SEO, topical authority, E-E-A-T signals — these still determine who enters the citation pool.

Add: Answer-first formatting. Every section should open with a direct answer to its heading question, ideally in the first 40 to 60 words. AI retrieval systems extract these opening passages. If the answer is buried in paragraph four, it won't be extracted cleanly.

Add: External citations within the content. Not just a sources section at the bottom. Name the study, name the organization, and link to the primary source mid-paragraph. According to the Princeton GEO paper, this single change has the highest measured impact on AI citation visibility of any content modification tested.

Add: A structured FAQ section with self-contained answers. Perplexity and ChatGPT surface FAQ-style answers frequently. Each answer should make sense without the surrounding article context.

Stop: Expecting schema markup to solve AI visibility. It handles structured SERP features. It does not change how AI systems retrieve or cite your content based on current evidence.

Stop: Optimizing purely for keywords. AI systems answer intent, not queries. Content that answers the underlying question thoroughly outperforms keyword-stuffed content in AI retrieval.

Add: Presence on multiple platforms. Perplexity indexes the open web. ChatGPT's web search feature does the same. llms.txt on your domain signals to AI crawlers what your site covers. Building entity signals across your author profiles, LinkedIn, and press mentions helps AI systems associate credibility with your content.

The 154.3 million US voice assistant users projected by eMarketer for 2025 represent a third search surface beyond Google and AI chatbots. Voice answers are typically AI-generated. The same content signals that help in Perplexity tend to help in voice results too.

---

Need help adapting your SEO strategy for AI search? We build content strategies that rank on Google and get cited in AI answers — including GEO audits, structured content frameworks, and llms.txt implementation. See what we do or get in touch to discuss your situation.

---

Frequently Asked Questions

What is generative engine optimization (GEO)?

GEO is the practice of optimizing content for visibility in AI-generated answers from systems like Google AI Overviews, ChatGPT, and Perplexity. A Princeton University study (KDD 2024) found that adding statistics improved AI citation visibility by 41%, and citing external sources improved it by up to 115% for lower-ranked content.

How much has organic CTR declined due to AI Overviews?

Organic CTR fell 61% for informational queries when AI Overviews appear, based on Seer Interactive research covering 25.1 million impressions across 42 organizations (Seer Interactive, November 2025). Paid CTR fell 68% on the same queries. Even queries without AI Overviews saw a 41% organic CTR decline in the same study.

Does schema markup improve AI citation rates?

Schema markup alone does not significantly improve AI citation rates. An Ahrefs study tracking 1,885 pages that added JSON-LD schema markup found changes of only a few percentage points in Google AI Overviews, ChatGPT, and Gemini citations, none of which were statistically significant (Ahrefs, 2026). Schema remains valuable for traditional rich results.

What type of content performs best in AI search?

Content with external citations, embedded statistics, and direct quotations performs best in AI search results. Princeton University research (KDD 2024, GEO-bench, 10,000 queries) found that citing external sources improved AI citation visibility by up to 115%, adding statistics improved it by 41%, and adding direct quotations improved it by 28% (Princeton/ACM).

Should businesses stop investing in SEO because of AI search?

No. 52% of Google AI Overview citations come from top-10 organic results, and the number-one ranked result has a 58% probability of being cited (Originality.AI, 2025). SEO still determines who enters the AI citation pool. The new requirement is adapting content structure so machines can extract and summarize your answers cleanly.

Conclusion

The AI search shift has already happened. Zero-click searches now represent over half of all Google queries. AI Overviews cover up to a quarter of all queries and cut organic CTR by 61%. ChatGPT and Perplexity have absorbed hundreds of millions of monthly searches.

The answer isn't to stop doing SEO. It's to do SEO and GEO together. Ranking still earns you entry into AI citation pools. Content structure — external citations, embedded statistics, answer-first formatting, self-contained FAQ answers — determines whether you get pulled from those pools into the generated answer.

Schema markup is not the lever. Your traditional rankings are. Your content structure is. Both are addressable right now, with existing content.