Google’s AI Overviews Didn’t ‘Steal’ Your Clicks — The New Metric Brands Are Quietly Buying Instead (and why it rewires what “going viral” even means in 2026)
AI Overviews don’t have to “steal” traffic to break the old web bargain. When Google answers first and links second, brands start paying for inclusion, recall, and authority—not clicks.
By TheMurrow Editorial
May 22, 2026
Key Points
- 1Expect CTR declines when AI Overviews trigger: Pew found 8% clicks with summaries vs. 15% without—nearly halving behavior.
- 2Plan for SEO’s old reward to weaken: Ahrefs reports the #1 result averages 58% lower CTR when an AI Overview appears.
- 3Optimize for inclusion and authority, not just visits: citations shape perception even when they don’t convert into traffic or revenue.
A few years ago, a search result was a promise: click a link and you’ll find the answer. Now Google increasingly delivers the answer before you decide whether the open web deserves your time.
That shift has a name—AI Overviews—and it has become the most contested piece of real estate on the internet. Publishers call it a siphon. Google calls it help. Users mostly treat it as a shortcut.
The uncomfortable part is that everyone can be right at once. The data we have—still incomplete—suggests people click less when AI summaries appear, even when those summaries cite the very sites that did the work.
When the answer is synthesized on the results page, the open web becomes optional.
— — TheMurrow
The question for 2026 isn’t whether AI Overviews change search behavior. They do. The question is what replaces the old bargain—publish content, earn clicks—when Google can quote the web without sending the web much traffic back.
What AI Overviews are—mechanically, not rhetorically
Google’s AI Overviews are AI-generated summaries that appear at or near the top of the search results page, along with citations and links to external pages that support the summary. They are not “featured snippets 2.0” so much as a new container that absorbs the user’s attention before traditional results even enter the conversation.
A crucial detail for publishers: Google treats the entire AI Overview module as one position in the results. In Google Search Console, that means all the links inside the module “inherit” that same position for reporting purposes. Google documents that behavior in Search Console support guidance, and it matters because it reshapes how performance looks in analytics. A link can appear in an AI Overview, generate impressions, and even earn clicks—yet the reporting can blur what happened inside the module versus what happened in classic “blue links.” Google also notes that clicks on those external links do count as clicks in Search Console.
A crucial detail for publishers: Google treats the entire AI Overview module as one position in the results. In Google Search Console, that means all the links inside the module “inherit” that same position for reporting purposes. Google documents that behavior in Search Console support guidance, and it matters because it reshapes how performance looks in analytics. A link can appear in an AI Overview, generate impressions, and even earn clicks—yet the reporting can blur what happened inside the module versus what happened in classic “blue links.” Google also notes that clicks on those external links do count as clicks in Search Console.
From SGE to AI Overviews: what changed
AI Overviews are the public-facing successor to Google’s earlier experiment, the Search Generative Experience (SGE). The modern rollout began in earnest during Google I/O week.
- May 14, 2024: Google announced AI Overviews and began rolling them out broadly in the U.S., with a stated ambition (reported at the time) to reach over a billion users by the end of 2024.
- 2025: Google expanded AI surfaces further, including AI Mode, a more interactive, chat-like search experience (timing varied by market and testing status).
- 2026: AI-powered search is no longer framed as a side experiment; mainstream coverage describes it as part of Google’s “next chapter” for Search.
None of those dates prove harm or benefit. They do establish a reality: for many queries, the first “result” is no longer a link. It’s Google speaking in paragraphs.
- May 14, 2024: Google announced AI Overviews and began rolling them out broadly in the U.S., with a stated ambition (reported at the time) to reach over a billion users by the end of 2024.
- 2025: Google expanded AI surfaces further, including AI Mode, a more interactive, chat-like search experience (timing varied by market and testing status).
- 2026: AI-powered search is no longer framed as a side experiment; mainstream coverage describes it as part of Google’s “next chapter” for Search.
None of those dates prove harm or benefit. They do establish a reality: for many queries, the first “result” is no longer a link. It’s Google speaking in paragraphs.
The core dispute: “stolen clicks” vs. a changing definition of value
Arguments about AI Overviews often collapse into moral language—“theft,” “scraping,” “appropriation.” Strong words, but the more useful debate has two layers.
First: Do AI Overviews reduce click-through to publishers? Second: if clicks decline, what becomes the scarce commodity that creators and brands can still sell?
The open web’s business model has long been anchored to a measurable behavior: the click. A publisher invests in reporting, editors refine it, SEO teams package it, and search sends traffic back. Display advertising, subscriptions, affiliate revenue—many flows depend on that visit.
AI Overviews don’t remove citations; they often add them. The problem is that the user may not need to click the citations to feel satisfied. The citation becomes a footnote, not a doorway.
First: Do AI Overviews reduce click-through to publishers? Second: if clicks decline, what becomes the scarce commodity that creators and brands can still sell?
The open web’s business model has long been anchored to a measurable behavior: the click. A publisher invests in reporting, editors refine it, SEO teams package it, and search sends traffic back. Display advertising, subscriptions, affiliate revenue—many flows depend on that visit.
AI Overviews don’t remove citations; they often add them. The problem is that the user may not need to click the citations to feel satisfied. The citation becomes a footnote, not a doorway.
Citations are not distribution. A link that doesn’t get clicked is just a receipt.
— — TheMurrow
Why this dispute is so hard to settle with a single number
Even the cleanest metrics become slippery because AI Overviews don’t trigger uniformly. They appear more often on queries that signal complexity or explanation-seeking. That query mix matters: a person searching “how to remove coffee stains from carpet” behaves differently from someone searching “weather.”
So when publishers say, “My traffic dropped,” they may be describing something real. When Google implies, “We still send traffic,” that may also be true. The fight is over the shape of the funnel—and who controls the moment the user decides they’ve learned enough.
So when publishers say, “My traffic dropped,” they may be describing something real. When Google implies, “We still send traffic,” that may also be true. The fight is over the shape of the funnel—and who controls the moment the user decides they’ve learned enough.
What the best behavioral evidence says: fewer clicks when summaries appear
For all the heated rhetoric, one of the clearest public data points comes from the Pew Research Center. In a July 22, 2025 analysis using browsing data from U.S. panelists, Pew compared visits where an AI summary appeared with visits where it did not.
The headline finding is blunt:
- Users clicked a result 8% of the time when an AI summary appeared.
- Users clicked a result 15% of the time when an AI summary did not appear.
That is nearly a halving of click behavior in the presence of an AI summary. Pew also cautioned that summaries and interfaces can change over time—an important reminder that Google can tune the product faster than the public can measure it.
The headline finding is blunt:
- Users clicked a result 8% of the time when an AI summary appeared.
- Users clicked a result 15% of the time when an AI summary did not appear.
That is nearly a halving of click behavior in the presence of an AI summary. Pew also cautioned that summaries and interfaces can change over time—an important reminder that Google can tune the product faster than the public can measure it.
8% vs. 15%
Pew (July 2025) found users clicked results 8% of the time with an AI summary, versus 15% without—nearly halving click behavior.
What “8% vs. 15%” really means in practice
On the ground, that gap looks like a familiar phenomenon: satisficing. Users accept a good-enough answer and move on. Sometimes they refine the query. Sometimes they stop searching. Sometimes they remain inside Google’s surfaces rather than jumping outward.
For readers, the practical implication is subtle: AI Overviews are not only a convenience. They also act as a filter that narrows which sources earn the deeper engagement that comes from a full visit—time on page, newsletter signups, subscriptions, brand loyalty.
For publishers, it means the first problem is not ranking. The first problem is persuading users that the summary is incomplete.
For readers, the practical implication is subtle: AI Overviews are not only a convenience. They also act as a filter that narrows which sources earn the deeper engagement that comes from a full visit—time on page, newsletter signups, subscriptions, brand loyalty.
For publishers, it means the first problem is not ranking. The first problem is persuading users that the summary is incomplete.
SEO measurement: steep CTR drops when AI Overviews trigger—especially at #1
If Pew gives us a behavioral snapshot, SEO tooling studies try to map how the presence of AI Overviews correlates with click-through rates across rankings.
An updated Ahrefs study using December 2025 data (published in 2026) reported a stark relationship: when an AI Overview appeared, the #1 ranking page on those keywords saw a 58% lower average CTR. Positions 2–10 also dropped, though less dramatically.
That’s a meaningful number because it speaks directly to the old SEO bargain: reach #1 and you’ll harvest disproportionate traffic. AI Overviews weaken that reward.
An updated Ahrefs study using December 2025 data (published in 2026) reported a stark relationship: when an AI Overview appeared, the #1 ranking page on those keywords saw a 58% lower average CTR. Positions 2–10 also dropped, though less dramatically.
That’s a meaningful number because it speaks directly to the old SEO bargain: reach #1 and you’ll harvest disproportionate traffic. AI Overviews weaken that reward.
58%
Ahrefs (Dec 2025 data, published 2026) reported the #1 ranking page saw a 58% lower average CTR when an AI Overview appeared.
A nuance many people miss: AI Overview CTR can rise while publishers still lose
Search Engine Land reported in 2026 that one dataset showed “early signs of recovery” for AI Overview click-through itself—from 1.3% (Dec 2025) to 2.4% (Feb 2026).
That sounds encouraging until you ask: clicks to whom? A higher CTR on the module could mean more clicks on cited publishers, yes. It could also mean users are clicking within Google’s experiences or selecting different paths that still reduce overall visits to independent sites.
The studies also share a limitation: correlation is not causation. AI Overviews tend to appear on certain types of queries, and those queries might have different click patterns even without AI summaries. The best research either matches similar queries or uses natural experiments around rollouts, but public data is still catching up to how fast the interface changes.
That sounds encouraging until you ask: clicks to whom? A higher CTR on the module could mean more clicks on cited publishers, yes. It could also mean users are clicking within Google’s experiences or selecting different paths that still reduce overall visits to independent sites.
The studies also share a limitation: correlation is not causation. AI Overviews tend to appear on certain types of queries, and those queries might have different click patterns even without AI summaries. The best research either matches similar queries or uses natural experiments around rollouts, but public data is still catching up to how fast the interface changes.
1.3% → 2.4%
One dataset cited by Search Engine Land (2026) showed AI Overview CTR rising from 1.3% (Dec 2025) to 2.4% (Feb 2026)—without guaranteeing publisher traffic recovery.
The harsh truth: ‘ranking’ is no longer synonymous with ‘receiving.’
— — TheMurrow
Measurement is now part of the problem: Search Console can’t cleanly separate AI Overview traffic
Publishers want a simple answer: “How much traffic did AI Overviews take from me?” Google’s current tooling doesn’t make that easy.
Google says AI Overviews follow standard Search Console rules for impressions, clicks, and position, and that the AI Overview module is treated as a single position with links inheriting it. Yet Search Console does not offer a clean, built-in segmentation that lets a publisher isolate AI Overview performance from classic results performance.
That absence turns analytics into inference. Teams compare time ranges. They look for query-level drops. They triangulate using third-party rank trackers. None of those approaches are perfectly reliable when the interface, triggers, and layouts can shift week to week.
Google says AI Overviews follow standard Search Console rules for impressions, clicks, and position, and that the AI Overview module is treated as a single position with links inheriting it. Yet Search Console does not offer a clean, built-in segmentation that lets a publisher isolate AI Overview performance from classic results performance.
That absence turns analytics into inference. Teams compare time ranges. They look for query-level drops. They triangulate using third-party rank trackers. None of those approaches are perfectly reliable when the interface, triggers, and layouts can shift week to week.
Practical takeaway: what publishers and brands can measure now
Until Google provides clearer reporting, teams can still do useful work:
- Query auditing: Identify which queries show the steepest CTR decline and check whether AI Overviews are now appearing for them.
- Content mapping: Track which pages are most frequently cited in Overviews and compare their click trends against similar pages not cited.
- On-site conversion tuning: If raw visits fall, improve conversion per visit—newsletter prompts, membership funnels, and clear “next step” pathways.
These aren’t glamorous tactics. They are survival tactics. When distribution becomes uncertain, efficiency matters more.
- Query auditing: Identify which queries show the steepest CTR decline and check whether AI Overviews are now appearing for them.
- Content mapping: Track which pages are most frequently cited in Overviews and compare their click trends against similar pages not cited.
- On-site conversion tuning: If raw visits fall, improve conversion per visit—newsletter prompts, membership funnels, and clear “next step” pathways.
These aren’t glamorous tactics. They are survival tactics. When distribution becomes uncertain, efficiency matters more.
What you can measure right now (even without clean GSC segmentation)
- ✓Audit queries with the biggest CTR drops and verify whether AI Overviews are triggering
- ✓Track which pages are cited in Overviews and compare click trends vs. similar non-cited pages
- ✓Improve conversion per visit with stronger newsletter prompts, membership funnels, and clearer next steps
The emerging academic lens: quality, fidelity, and the economics of being cited
In May 2026, a large-scale measurement study posted to arXiv examined AI Overviews across 55,393 trending queries, collected between March 13 and April 21, 2026. The study looked at activation patterns, source quality, “claim fidelity,” and publisher impact.
One finding lands like a quiet alarm bell for media economics: well over half of the pages cited in AI Overviews carry display advertising. If AI Overviews reduce clicks—and the evidence from Pew and Ahrefs suggests they often do—then the sites most likely to be cited are also the sites most likely to feel revenue pain when users don’t arrive.
This creates a strange inversion. The web’s informational labor is recognized (you are cited), but the web’s business model is weakened (you are not visited). Citation becomes a kind of prestige without payment.
One finding lands like a quiet alarm bell for media economics: well over half of the pages cited in AI Overviews carry display advertising. If AI Overviews reduce clicks—and the evidence from Pew and Ahrefs suggests they often do—then the sites most likely to be cited are also the sites most likely to feel revenue pain when users don’t arrive.
This creates a strange inversion. The web’s informational labor is recognized (you are cited), but the web’s business model is weakened (you are not visited). Citation becomes a kind of prestige without payment.
55,393
A May 2026 arXiv study examined AI Overviews across 55,393 trending queries collected March 13–April 21, 2026.
Multiple perspectives: users, Google, publishers
A fair reading includes all three constituencies:
- Users gain speed and clarity, especially on complex questions where ten blue links feel like homework.
- Google reduces friction and keeps users satisfied, which is its core incentive.
- Publishers face a conversion problem: they provide inputs to an answer that competes with their own article.
None of that proves malicious intent. It does show a structural tension: the better Google gets at summarizing, the less the average user needs to leave Google.
- Users gain speed and clarity, especially on complex questions where ten blue links feel like homework.
- Google reduces friction and keeps users satisfied, which is its core incentive.
- Publishers face a conversion problem: they provide inputs to an answer that competes with their own article.
None of that proves malicious intent. It does show a structural tension: the better Google gets at summarizing, the less the average user needs to leave Google.
What replaces the click: brand recall, authority, and “earned inclusion” in AI answers
If clicks become scarcer, the question becomes: what should publishers, experts, and brands optimize for?
The uncomfortable answer is that search is drifting from traffic acquisition toward influence acquisition. Being cited in an AI Overview may shape perception even when it doesn’t drive a visit.
That matters for:
- Public health guidance
- Consumer purchase decisions
- Financial education
- Political and civic information
If AI Overviews become the first draft of many people’s understanding, then the fight is as much about who gets to define the summary as about who receives the click.
The uncomfortable answer is that search is drifting from traffic acquisition toward influence acquisition. Being cited in an AI Overview may shape perception even when it doesn’t drive a visit.
That matters for:
- Public health guidance
- Consumer purchase decisions
- Financial education
- Political and civic information
If AI Overviews become the first draft of many people’s understanding, then the fight is as much about who gets to define the summary as about who receives the click.
A real-world scenario you’ve likely lived
You search for advice—say, how to negotiate rent, choose a mortgage, treat a minor injury, or evaluate a travel warning. You read the AI Overview, scan two citations, and stop.
In the old model, you might have opened three tabs and compared them. In the new model, Google compares them for you. Your attention moves from “which article do I trust?” to “does this summary feel plausible?”
For publishers, that means authority signals—clear authorship, transparent sourcing, demonstrable expertise—matter not only for rankings, but for being selected as a citation in the first place.
In the old model, you might have opened three tabs and compared them. In the new model, Google compares them for you. Your attention moves from “which article do I trust?” to “does this summary feel plausible?”
For publishers, that means authority signals—clear authorship, transparent sourcing, demonstrable expertise—matter not only for rankings, but for being selected as a citation in the first place.
In AI-mediated search, the first battle is not for the click. It’s for the sentence.
— — TheMurrow
Practical implications for creators and organizations
Without inventing new metrics, a few grounded implications follow from the research:
- Expect lower CTR on queries where AI Overviews appear (Pew’s 8% vs. 15%; Ahrefs’ 58% drop for #1).
- Treat being cited as a partial win—but not a financial one unless it produces visits or downstream conversion.
- Invest in content that gives users a reason to click: original reporting, tools, checklists, data, or perspective that can’t be compressed into a generic paragraph.
- Expect lower CTR on queries where AI Overviews appear (Pew’s 8% vs. 15%; Ahrefs’ 58% drop for #1).
- Treat being cited as a partial win—but not a financial one unless it produces visits or downstream conversion.
- Invest in content that gives users a reason to click: original reporting, tools, checklists, data, or perspective that can’t be compressed into a generic paragraph.
Key Insight
Search is drifting from traffic acquisition to influence acquisition: inclusion in AI summaries can shape perception even when it doesn’t produce a visit.
Where this leaves readers: convenience, trust, and the cost of frictionless knowledge
Readers are not villains for wanting quick answers. The web trained everyone to expect speed. AI Overviews deliver speed—sometimes with real utility.
The risk is epistemic. When answers arrive pre-digested, fewer people practice the habits that keep information honest: reading primary sources, comparing viewpoints, noticing uncertainty, and seeing who benefits from a framing. AI Overviews cite sources, but citations are not the same as scrutiny.
The more search becomes a single narrative voice, the more it matters which sources feed that voice—and how accurately they are represented. The 2026 arXiv work’s attention to “claim fidelity” points toward the right question: not only “Is the summary helpful?” but “Is the summary faithful?”
A web that survives this transition will likely look different: fewer commodity explainers, more distinctive work; fewer anonymous listicles, more accountable expertise; fewer businesses built entirely on search arbitrage.
That is not nostalgia. It’s adaptation under pressure. AI Overviews did not end the open web. They did change the price of admission.
The risk is epistemic. When answers arrive pre-digested, fewer people practice the habits that keep information honest: reading primary sources, comparing viewpoints, noticing uncertainty, and seeing who benefits from a framing. AI Overviews cite sources, but citations are not the same as scrutiny.
The more search becomes a single narrative voice, the more it matters which sources feed that voice—and how accurately they are represented. The 2026 arXiv work’s attention to “claim fidelity” points toward the right question: not only “Is the summary helpful?” but “Is the summary faithful?”
A web that survives this transition will likely look different: fewer commodity explainers, more distinctive work; fewer anonymous listicles, more accountable expertise; fewer businesses built entirely on search arbitrage.
That is not nostalgia. It’s adaptation under pressure. AI Overviews did not end the open web. They did change the price of admission.
T
About the Author
TheMurrow Editorial is a writer for TheMurrow covering trends.
Frequently Asked Questions
What are Google AI Overviews, exactly?
AI Overviews are AI-generated summaries shown at or near the top of Google Search results, often with citations/links to external web pages. They aim to answer queries directly on the results page, reducing the need to click multiple results. Google treats the AI Overview module as a single results position for reporting purposes in Search Console.
When did AI Overviews roll out to most users?
Google announced and began broadly rolling out AI Overviews in the U.S. on May 14, 2024 during Google I/O week. Reporting at the time described Google’s intention to reach over a billion users by the end of 2024. Expansion and new AI search experiences continued through 2025 and into 2026.
Do AI Overviews reduce clicks to websites?
Evidence points to “yes” in many cases. Pew Research Center (July 2025) found users clicked results 8% of the time when an AI summary appeared, versus 15% when it didn’t. Ahrefs (Dec 2025 data, published 2026) reported the #1 ranking page saw 58% lower average CTR when an AI Overview was present.
Do clicks from AI Overview citations count in Google Search Console?
Yes. Google states that clicks on external links inside AI Overviews count as clicks, and impressions/position follow standard Search Console rules. A key caveat: Search Console does not currently provide a clean way to segment AI Overview performance separately from classic blue-link performance, making analysis harder.
If AI Overviews cite publishers, why are publishers still upset?
Because a citation doesn’t guarantee a visit. AI Overviews can satisfy the user directly on Google, so the user may not click through even when the publisher’s page is cited. Since many publishers rely on visits for ad revenue, subscriptions, or affiliate income, fewer clicks can translate into real financial harm despite “visibility.”
What should publishers and brands do if search clicks decline?
Focus on what AI can’t fully replace: original reporting, unique data, tools, and accountable expertise. Audit queries where Overviews appear, monitor which pages get cited, and improve on-site conversion so fewer visits still produce outcomes (subscriptions, signups, purchases).
