Your top-ranked page used to earn the lion’s share of clicks. Now Google answers the question above it, and most users never scroll. AI Overviews have rewritten how organic traffic flows from search results, and the numbers are not subtle. In my experience auditing portfolios through 2025 and into 2026, no single SERP feature has shifted CTR distributions this aggressively since the introduction of Featured Snippets.
This guide explains what AI Overviews actually are, how Google picks the sources it cites, what the traffic impact looks like across credible studies, when AI Overviews appear (and when they don’t), and how to position your content for citation. No tactical setup walkthroughs — just the concepts you need to make sane decisions about content investment, audit priorities, and where to push back when stakeholders demand reactive rewrites of pages that aren’t actually exposed.
What Google AI Overviews Actually Are
AI Overviews are Google-generated summaries that appear above traditional organic results for certain queries. Powered by a version of Gemini fine-tuned for Search, they synthesize information from multiple sources, present a direct answer, and cite the pages they drew from. Google launched the feature in the U.S. in May 2024 and expanded globally through 2025.
Importantly, AI Overviews don’t replace the traditional 10 blue links. Instead, they sit above them. However, when an Overview appears, it consumes the vertical space users see first — sometimes the entire above-the-fold area on mobile. As a result, the click economy of the page changes substantially.
According to Google’s own Search Central guidance, AI Overviews draw from the same index that powers traditional results. There is no separate ranking system. The same E-E-A-T signals, technical health, and topical depth that influence organic ranking also influence citation eligibility.
Key clarification: AI Overviews are a presentation layer on top of regular search. Pages must already rank (typically in the top 10) to qualify for citation.
How AI Overviews Pick Sources (and What That Means for SEO)
Source selection in AI Overviews is not a separate algorithm with documented inputs. Instead, Google chooses citations from pages that already rank well for the query and that match specific structural and content patterns. Several patterns emerge consistently across third-party studies.
First, ranking position matters. Specifically, pages in the top 10 are far more likely to be cited, with citation probability rising as ranking improves. Second, content structure influences eligibility. For example, pages with clear lists, tables, FAQs, and definition-style passages get cited more often than long unstructured prose. Third, recency plays a role — similarly, recent analysis from The Digital Bloom found that roughly 85% of AI Overview citations came from content published in the past few years.
However, ranking in the top 10 no longer guarantees citation the way it once did. A 2026 ALM Corp analysis reported that the share of AI Overview citations drawn from top-10 results dropped from 76% to 38% over twelve months, indicating Google is pulling from a wider citation pool — including Reddit, YouTube, LinkedIn, and forum-style content.
When I tested this on a 40-site portfolio across multiple verticals in late 2025, the pattern held: pages cited in AI Overviews almost always met three conditions simultaneously. They ranked in the top 10 for the underlying query, they answered the question within the first 100 words, and they had at least one structured element (list, table, or definition block) that the summary could lift cleanly. Pages missing any one of those three rarely earned citation, even when their domain authority was higher than competitors.
The “passage extraction” reality
AI Overviews don’t cite pages — they cite passages. Specifically, the summary lifts a sentence or short paragraph that directly answers part of the query, and the citation badge links to the source page anchored at or near that passage. As a result, a 4,000-word pillar article and a 600-word focused answer page can compete for the same citation slot. The deciding factor is which page contains the cleanest, most extractable answer for that specific sub-question.
Essentially, this changes content strategy at a structural level. For instance, a page can rank in position 2 organically and still lose every AIO citation to a position-6 page that happens to have a better-formatted FAQ block. Conversely, a page can sit in position 8 and become the primary AIO citation if its answer paragraph is the cleanest match. In my experience, this is why content audits now need to evaluate passages, not just pages.
What gets cited most
Across a Search Engine Land study of AI engine citations, the top-cited domains looked remarkably similar across ChatGPT, Gemini, Perplexity, and Google AI Overviews:
- Reddit — first-person user discussions, comparison threads, “best of” lists
- YouTube — transcripts and descriptions of how-to and review content
- LinkedIn — professional opinions, B2B commentary
- Wikipedia — definitions, neutral overviews, historical context
- Forbes and similar editorial sites — opinion and trend articles
The pattern is clear: AI systems prefer sources that look like authentic, attributable human voices over generic marketing copy. Consequently, for independent publishers, this means a topical hub built on demonstrable expertise outperforms a thin product page nearly every time. To dig deeper into the signals Google uses to assess credibility, see my guide to E-E-A-T signals.
The Traffic Impact — What the Data Shows
The headline numbers vary by methodology, but the direction is consistent. Every credible study published between mid-2024 and early 2026 reports a meaningful CTR drop when AI Overviews appear above organic results. Below is a summary of the most-cited research.
| Study | Sample | CTR Impact (with AI Overview) |
|---|---|---|
| Pew Research (Jul 2025) | 68,879 searches, 900 U.S. adults | Clicks on traditional results fell from 15% to 8% |
| Ahrefs (initial, Apr 2025) | 300,000 keywords | 34.5% CTR drop for position-one organic result |
| Ahrefs (updated, Dec 2025) | Top-ranking pages | 58% lower average CTR for top-ranked page |
| Seer Interactive (Sep 2025) | Query sample analysis | Organic CTR dropped 61% (1.76% → 0.61%) |
| Amsive | 700,000 keywords | Average 15.49% CTR decline; up to 37% with Featured Snippets present |
| Authoritas | Case studies | ~79% drop for top organic link when AIO present |
| BrightEdge | Industry tracking | 30% CTR drop overall; impressions up 49% |
The Pew Research study is the most methodologically rigorous because it observed real user behavior rather than estimating from rank-tracking tools. Specifically, its finding — clicks on links within AI summaries themselves occur in only 1% of visits — confirms what most SEO teams already see in Search Console. In other words, when Google answers the question, the user often stops there.
Google publicly disputed the Pew findings as “flawed methodology and skewed queryset,” though notably the company has not released its own clickstream data. In my experience working with mid-size publisher portfolios through 2025, the Pew numbers track closely with what I see in raw Search Console exports for informational query clusters.
The cited-brand offset
However, one finding pushes against the doom narrative. Specifically, brands cited inside an AI Overview earn approximately 35% more organic clicks and 91% more paid clicks compared to ranking outside the Overview. As a result, citation visibility provides partial compensation — though it requires being chosen as a source, which is a much smaller pool than the old top-10.
Why the numbers diverge so widely
The studies above report CTR drops ranging from 15% to 79%, which can look contradictory at first glance. The variance is methodological, not random. Specifically:
- Pew Research measured real user clickstream data from a panel of consenting U.S. adults, capturing actual behavior across all query types
- Ahrefs used rank-tracking + Search Console data, weighting toward higher-volume keyword sets
- Authoritas reported case-study CTRs that skew higher because they focus on queries where AIOs reliably appear
- Amsive averaged across 700,000 keywords including many low-AIO-presence terms, dragging the headline number down
- BrightEdge tracked aggregated impression and click data over time, capturing the broader market effect
Therefore, when communicating with stakeholders, anchor on the methodology that matches your situation. A B2B SaaS team with a small set of high-intent keywords should reference the Authoritas or Ahrefs numbers. A large publisher with broad informational coverage should anchor on Pew and Amsive figures. Citing the wrong study to the wrong audience erodes credibility fast.
When AI Overviews Appear vs When They Don’t
AI Overviews are not universal. They appear selectively based on query intent, vertical, and perceived risk. Understanding the trigger patterns matters because not every traffic source is equally exposed. An Ahrefs analysis of 146 million SERPs identified 86 distinct factors that correlate with AI Overview appearance.
Intent patterns (late 2025 data)
- Informational queries trigger AI Overviews most aggressively — early 2025 data showed up to 98% trigger rate, settling to roughly 57% by late 2025 as Google rebalanced
- Commercial investigation queries trigger AIOs around 40% of the time, with the share rising through 2025
- Navigational queries grew from under 1% trigger rate in January 2025 to over 10% by October 2025
- Transactional queries — buy intent, checkout, pricing decisions — still rarely trigger AIOs because Google protects ad revenue and conversion-stage real estate
Industry exposure
Furthermore, according to a Semrush AI Overviews study, the verticals with the highest AIO share include:
- Science — 43.6% of queries
- Health — 43.0%
- Pets and animals — 36.8%
- People and society — 35.3%
- Computers and electronics — ~18% but rising
Healthcare informational queries hit 83.6% AIO trigger rate. Meanwhile, ecommerce queries declined to roughly 18.5% AIO presence as Google protects transactional searches. The takeaway: a SaaS pricing page or product comparison is far less exposed than a “what is” or “how does X work” article in the same niche.
How to Position Content to Be Cited (Concepts, Not Tactics)
The goal has shifted. Earning a top-3 organic ranking is no longer the entire game — being chosen as a source in the AI Overview is the new ceiling. The principles below are conceptual; for tactical execution, see my guide to AI search optimization (AEO).
1. Answer the question before you sell anything
Basically, AI Overviews extract concise, factual statements. For example, pages that lead with a direct definition or answer — typically in the first paragraph — get cited far more than pages that bury the answer under a sales narrative. Likewise, the “inverted pyramid” structure from journalism works well here.
2. Structure for extraction
Lists, tables, FAQs, and clearly bounded definitions are easier for the underlying LLM to lift cleanly. Therefore, a well-formatted comparison table or numbered procedure tends to be cited over the equivalent information buried in prose. This doesn’t mean writing only in bullets — it means making sure key facts have a structurally clean home on the page. My on-page SEO guide covers the formatting patterns in depth.
3. Demonstrate first-hand experience
E-E-A-T started as a quality rater guideline. However, now it functions as a citation filter. Specifically, AI Overviews show a measurable preference for content with author attribution, original data, tested claims, and clear methodology. In contrast, generic AI-generated summaries of other people’s research rarely get cited.
4. Treat structured data as a credibility ladder
Schema markup doesn’t directly cause citation. However, it helps Google parse and trust your content faster. For instance, Article, FAQ, HowTo, and Organization schema work together to clarify authorship, question-answer pairs, and entity relationships. For implementation patterns, see the schema markup guide.
5. Reverse-engineer query intent
Importantly, not every keyword deserves the same treatment. For example, informational queries in your topic cluster need answer-first structure; commercial queries need decision frameworks; meanwhile, transactional queries need clean product information. The keyword research guide for 2026 covers how to map intent to format under the new conditions.
6. Build topical density, not single-page authority
AI Overviews tend to cite sites that demonstrate coverage across a topic, not just one strong page. In other words, a domain with twelve well-connected articles on a subject signals expertise more strongly than a single 5,000-word monolith on the same topic. This favors hub-and-spoke architectures over pillar-only strategies. Internal linking between related articles helps both human readers and Google’s understanding of your topical scope.
When NOT to chase AI Overview citation
AIO citation is not the right goal for every page. Specifically, transactional landing pages, lead capture pages, and pricing pages typically should not aim for citation — they should convert traffic, not feed answers. Likewise, if your category sees less than 15% AIO presence, your effort is better spent on traditional ranking signals and conversion. Honest assessment of where AIOs actually appear in your niche prevents wasted optimization work.
Additionally, some queries are actively dangerous to chase. Medical, legal, and financial queries where Google heavily filters for credentialed sources will not cite a marketing blog regardless of content quality. In those YMYL verticals, a small consultancy is better served by guest posts and contributions to established publications than by trying to win citation slots from authority sites like Mayo Clinic or Investopedia.
What This Means for Different Industries
The CTR impact is not evenly distributed. As a result, strategy needs to differ by vertical. Below is a practical framing of how exposure varies.
| Industry | AIO Exposure | Practical Implication |
|---|---|---|
| Health & medical | Very high (43%+) | YMYL constraints + heavy AIO triggering = citation strategy critical; lean on credentialed authorship |
| Science & education | Very high (43%+) | Definitions and explanations dominate citations; structured content wins |
| SaaS & B2B | Medium (15-25%) | Top-of-funnel content exposed; comparison and pricing pages largely protected |
| Ecommerce | Low-medium (~18%) | Product and category pages mostly safe; informational blog content exposed |
| Local services | Low | Local pack and Map results take priority; AIO impact minimal |
| Finance (YMYL) | High | Heavy citation preference for established institutional sources; new entrants struggle |
| Travel | Medium-high | Destination guides and itinerary content cited; booking pages protected |
For instance, if you run a SaaS site, your pricing page is largely safe. However, your “what is [category]” pillar page is not. Likewise, if you publish medical content, you face both YMYL ranking constraints and aggressive AIO triggering — therefore, citation strategy moves from optional to existential.
What the data doesn’t tell you
The aggregated industry numbers conceal significant intra-industry variance. For instance, two sites in the same SaaS vertical can see wildly different AIO exposure depending on their content mix. A site heavy on “how does [feature] work” articles will face much higher AIO pressure than a competitor focused on “[product] vs [product]” comparison content. Therefore, before benchmarking against industry averages, audit your own query mix in Search Console — specifically, segment by query intent and measure AIO presence per segment.
Similarly, geography matters. AI Overviews rolled out unevenly across markets through 2025, and rollout in the EU was delayed by regulatory considerations. Sites targeting European audiences may see lower AIO presence than U.S. benchmarks suggest, though the gap is closing. International publishers should track AIO presence per market rather than assuming uniform exposure.
Common Misconceptions About AI Overviews
Several persistent myths about AI Overviews lead teams toward bad decisions. Below are the ones I correct most often when reviewing client strategy.
- “AI Overviews kill organic traffic universally.” They don’t. Transactional queries, branded queries, navigational queries, and most ecommerce SERPs remain largely AIO-free. The damage concentrates on informational content.
- “You can opt out and protect yourself.” You can opt out via
nosnippetor thedata-nosnippetattribute, but this also removes you from traditional snippets and severely cuts visibility. The trade-off is almost always net-negative. - “Schema markup forces citation.” It doesn’t. Schema helps Google parse content but isn’t a citation trigger by itself. Many heavily cited pages have minimal schema; many fully marked-up pages are never cited.
- “Long-form content gets cited more.” Not consistently. Citations frequently come from concise paragraphs and lists inside longer pages — but also from short, focused answer pages. Structure matters more than length.
- “AI Overviews will disappear or be regulated away.” Unlikely in any meaningful timeframe. Google has invested too heavily, and regulatory action would target ad-market dominance long before AIO presentation. Plan for permanence.
- “If you’re not cited, you’re irrelevant.” False. Plenty of pages still earn substantial traffic from positions 2-5 on queries where AIOs don’t trigger, or where users still scroll past the summary for deeper content.
Treating AIO impact as a calibration problem — not an extinction event — leads to better resource allocation. The portfolios I see succeeding are the ones that audit their content by AIO exposure level and apply different strategies to each segment. Conversely, the portfolios that struggle treat AI Overviews as a single monolithic threat and rewrite every page the same way, burning budget on transactional pages that were never going to be cited in the first place.
Bottom Line
Google AI Overviews have permanently changed CTR distributions for informational queries, with credible studies reporting drops between 15% and 79% depending on methodology and query type. The Pew Research finding of clicks falling from 15% to 8% when AIOs appear is the most defensible single data point, while the Ahrefs 58% drop for top-ranked pages captures the upper bound for high-traffic queries.
However, the impact is far from uniform. Health, science, and informational verticals face existential pressure; transactional, ecommerce, and local search remain largely intact. Therefore, the right response is not panic but segmentation: audit your content by AIO exposure, prioritize citation-friendly structure where AIOs trigger, and double down on conversion fundamentals where they don’t.
The bigger shift is conceptual. Ranking position one is no longer the ceiling. Being chosen as a source in the AI Overview is. That means the principles that have always mattered — clear answers, structured information, demonstrable expertise, fast and accessible pages — now matter more, not less. For the foundational work that supports both ranking and citation, start with a clean technical SEO audit, healthy Core Web Vitals, and a strategic approach to link building. The AI Overview era rewards the fundamentals, just with higher stakes.
