We Rank #1 and Get Zero Clicks: What AI Overviews Did to Search — and What It Means for Your Pipeline

A confession from our own search console

I'll start with our own numbers, because they're more honest than anyone else's, and because they make the abstraction concrete.

Last week, one of our highest-intent pages — a service page targeting a commercial query a buyer types when they're ready to spend money — sat at position 6.0 on Google. It collected 887 impressions. It earned zero clicks. Not a low click-through rate. Zero. A second page, ranking at position 5.77 on a strong commercial term, did the same thing: hundreds of impressions, page-one position, no clicks at all.

For most of the last fifteen years, those numbers would have been a contradiction. Page one with zero clicks didn't happen — if you ranked, you got traffic, in proportions you could roughly predict. It happens now, every week, and not because the pages are bad. Our behavioral analytics show clean, engaged reading once people actually arrive: no rage clicks, healthy scroll depth, real time on page. The clicks aren't being lost on the page. They're being lost before the page, in the search result itself.

This is the part most leaders haven't internalized yet, because it hides behind reassuring metrics. Impressions are up. Rankings are holding. Traffic is gone. If your dashboard tracks position and impressions — and most do — everything looks healthy right up until the moment you notice the pipeline has quietly thinned. The thing that broke isn't your SEO. It's the click. And because the break is invisible in the metrics most teams watch, it tends to get diagnosed late, usually as "our content isn't working anymore" — which sends people off optimizing the wrong thing entirely.

Citable: In 2026, a page can hold position one on Google and capture zero clicks. We see it in our own search console: pages at positions 5–6 on commercial queries, hundreds of impressions, no clicks. Ranking and traffic have decoupled — and dashboards that only track position and impressions will show "healthy" right up until the pipeline runs dry.

What actually changed: the answer moved above the links

The mechanism has a name now: the AI Overview. When you search Google, an AI-generated summary increasingly appears at the very top of the page, answering your question before any blue link. Google rolled its conversational AI Mode out broadly in the US in May 2025, and by 2026 it reached roughly a billion monthly users. The average AI Mode query runs about three times the length of a traditional search — people are asking fuller, more conversational questions and getting fuller, more conversational answers, with no obligation to click anything. The summary is no longer an experiment bolted onto search. For a large share of queries, it is the search.

The independent data on what that does to clicks is now unambiguous, and — importantly — it comes from sources with no incentive to exaggerate the damage.

Pew Research Center analyzed 68,879 real Google searches from 900 US adults — actual browsing behavior, not survey opinion about what people think they do. They found that when an AI summary appeared, only 8% of searches produced a click to any website, compared to 15% when no summary was present. Roughly half the clicks, gone. Only 1% of users clicked a link inside the summary itself — so the "well, at least we can get cited in the box" hope comes with a sobering ceiling on direct clicks. And users were measurably more likely to simply end their session after seeing an AI summary: 26% of the time, versus 16% without one. The answer satisfied them. There was nowhere they needed to go.

This is the context for understanding what we see in our own GSC data — and it's why we've written about the broader implications of Google's AI optimization guidance for mid-market strategy: the answer moved above the links, and the whole measurement system most teams use was built for a world where that didn't happen.

Citable: Pew Research found that when a Google AI summary appears, only 8% of searches result in a website click — versus 15% without one — and just 1% of users click a link inside the summary (Pew, July 2025). The AI Overview doesn't redirect the click. It replaces it.

The independent data agrees — from three different directions

What makes this real, rather than one alarmist study, is that independent researchers using completely different methodologies keep landing in the same place.

Ahrefs put a number on the ranking-level damage. Analyzing 300,000 keywords against Google Search Console data, they found that position-one click-through rate fell to about 1.6% when an AI Overview is present — down from 7.3% two years earlier, roughly a 78% collapse. Here's the detail that should stop anyone planning to wait this out: even without an AI Overview present, position-one CTR dropped to 3.9% over the same window — a 49% decline. Some of the erosion is AI Overviews directly. Some of it is users migrating to ChatGPT, Perplexity, and Claude before they ever reach Google. The behavior is shifting underneath the entire system, not just inside one Google feature — which means "ride it out until Google reverses the change" isn't a strategy, because Google reversing AI Overviews wouldn't bring the other half back.

Seer Interactive, tracking 3,119 search terms across 42 organizations and 25 million impressions, found organic click-through rate on informational queries with an AI Overview fell from 1.76% in June 2024 to 0.61% by September 2025 — a 61% decline in fifteen months. The same study found the inverse effect that points to the way out: when a brand was cited inside the AI Overview, its organic clicks ran 35% higher than for uncited queries (and paid clicks 91% higher). Being in the answer, it turns out, is worth more than being in the link list beneath it.

BrightEdge, reviewing a year of AI Overviews, measured Google search impressions rising 49% year-over-year while click-through rate fell roughly 30% over the same period — the exact signature of the rank-but-no-click pattern, at scale. (One honest caveat on BrightEdge: their client base skews heavily enterprise and Fortune 100, so read their absolute traffic figures as directional for the mid-market rather than precise.)

These aren't three versions of the same number. They're three independent methodologies — a behavioral panel, search-console aggregation, and enterprise tracking — converging on the same conclusion. That convergence is what moves this from "a thing SEO people are worried about" to "a structural change you should plan around."

Google says traffic is fine. Here's the part they're not saying.

Google's position is worth quoting directly, because you will hear it repeated — by vendors, by your own team, by anyone who'd rather not deal with this.

In August 2025, Liz Reid, VP of Google Search, stated that total organic click volume to websites has been "relatively stable" year-over-year, and that the clicks Google does send are higher quality. Search Advocate John Mueller has made the same "higher quality clicks" argument.

Take the claim seriously, then look at the denominator. BrightEdge measured Google search impressions rising 49% year-over-year while click-through rate fell. If click volume is "stable" while impressions explode by half, then the rate at which searches turn into visits has collapsed — which is exactly what every independent study shows. "Stable clicks" drawn from a vastly larger pool of searches is not good news. It's the same number of visitors pulled from a much bigger crowd that now mostly never leaves Google at all.

And the "higher quality" claim is, by design, unfalsifiable from the outside. There is no public dataset to check it against. It may even be true — fewer, more-committed visitors is a plausible outcome of the AI Overview filtering out casual clicks. But a business running content-led lead generation cannot pay its bills with an unverifiable quality improvement when 92% of searches now resolve without a click. If the top of your funnel just lost most of its volume, "the remaining trickle is higher quality" is not a strategy. It's a consolation prize, and you should recognize it as one when it's offered to you.

Citable: Google says organic clicks are "relatively stable" and "higher quality." But impressions rose 49% year-over-year (BrightEdge) while click-through rate fell — so "stable clicks" on a far larger base means the conversion rate collapsed. "Higher quality" is unfalsifiable from outside Google, and it doesn't refill a top-of-funnel that just lost most of its volume.

Ranking isn't dead. It just stopped being the finish line.

Here's where it gets counterintuitive, and where most of the panic-writing on this topic gets it wrong. The answer is not "SEO is dead, abandon your rankings." The relationship between ranking and visibility has changed, but it hasn't dissolved — and overcorrecting into "search doesn't matter anymore" is its own expensive mistake.

What's changed is the conversion event. For fifteen years, the event that mattered was the click: rank well, earn the click, capture the visitor, start the relationship. In an AI-Overview world, the event that increasingly matters is the citation — being the source the AI quotes when it composes its answer. When your brand is cited inside an AI Overview, Seer found organic clicks to your site run 35% higher than for uncited queries. Being the answer is the new being-on-page-one.

And ranking still feeds citation. BrightEdge's analysis found that a majority of AI Overview citations come from organically well-ranked pages — the share of citations drawn from strong organic positions has been climbing, not falling. So the work you've done to rank isn't wasted; it's repurposed. The question is just no longer "are we on page one?" It's "when the AI answers this question, are we the source it reaches for?"

Citable: The conversion event in search has moved from the click to the citation. Ranking still matters — it's a major input to whether you get cited — but the question is no longer "are we on page one?" It's "when the AI answers this question, are we the source it quotes?" Those are different games, and most companies are still optimizing for the one that's shrinking.

This reframe is liberating once it lands, because it turns an unwinnable fight into a winnable one. You are not trying to claw back clicks that the AI Overview is structurally designed to absorb — that's a fight against Google's interface, and you will lose it. You're trying to become the thing the AI reaches for, which is a content-and-architecture problem you can actually influence. It connects directly to the citation economy shift we've written about before: the winners in this environment aren't ranking harder, they're being cited more — and citation is earnable.

What actually earns the citation

The good news is that being cited isn't mystical, and it isn't a black box. Semrush ran one of the largest studies on this to date — 11,882 prompts across 337,785 URLs, examining what distinguishes pages that get cited in AI answers from pages that merely rank. The patterns are consistent and, refreshingly, unglamorous:

• Clear, summarizable writing was the strongest single predictor — pages with it were about 33% more likely to be cited.
• Demonstrated expertise and trust signals (real authorship, citations, evidence) — about 31% more likely.
• Question-and-answer structure — about 25% more likely.
• Clean section structure — about 23% more likely.
• Structured data — about 22% more likely.

None of that is a growth hack. It's a description of a genuinely good, well-organized, trustworthy page — one written to be quoted rather than to game a ranking algorithm. Which is the deeper point: AI Overviews have, almost by accident, made content quality and content architecture matter more than content volume. The page an AI can lift a clean, attributable answer from is the page that gets cited. The keyword-stuffed, padded-to-2,000-words page that ranked fine in 2022 is exactly what the AI skips over now, because there's no clean answer in it to quote.

In practice, this is what answer-engine optimization actually means, stripped of the jargon: write the answer to the question near the top, in a self-contained way an AI can quote without surrounding context. Use real question-and-answer structure. Show your expertise and cite your sources — AI systems are more likely to trust and quote content that itself cites credible sources. Add structured data so machines can parse what the page is about. Make the page legible to a model, not just to a crawler. This is why the response to AI search isn't a content tactic bolted onto your existing process — it's an architecture decision about how your knowledge is structured and how clearly your expertise is signaled. That's upstream of whether you're visible at all.

The opportunity hiding inside the threat

It would be a mistake to read all of this as pure loss. The same shift that's draining clicks from Google is opening a channel that barely existed two years ago: referral traffic from the AI tools themselves.

When someone asks ChatGPT, Perplexity, or Claude a question and the answer cites your company, some of those people click through to you. That traffic is small in volume today — BrightEdge puts AI search at under 1% of total referral traffic, and organic Google is still the primary driver for now. But the quality signal is striking. HockeyStack, analyzing 118 accounts, found that visitors arriving from large language models converted to a "hand-raiser" at a median rate of 5.24%, and that 86% of LLM-sourced leads were classified as high-intent. People who arrive because an AI recommended you tend to arrive pre-qualified — they've already been told, by a source they trust, that you're the answer. That's a warmer introduction than a cold organic click ever was.

The channel composition is worth knowing as you build for it: in HockeyStack's data, ChatGPT drove the large majority of LLM-referred sessions (around 82%), with Perplexity (around 12%) and Gemini (around 5%) splitting most of the rest. So "get cited by AI" practically means "get cited by ChatGPT first," with the others as meaningful secondary channels.

Citable: AI referral traffic is small but disproportionately valuable. HockeyStack found visitors arriving from AI assistants converted to hand-raisers at a median 5.24%, with 86% classified as high-intent. People who arrive because an AI recommended you arrive pre-qualified — which is why the redistribution underneath the click decline favors the companies that earn citations.

We see this in our own analytics. AI tools — Claude, Perplexity, ChatGPT — now show up as referral sources every week, sending visitors who read deeply and bounce less than the search-driven average. The volume is modest. The intent is not. This is the redistribution underneath the headline decline: top-of-funnel click volume is contracting, but a smaller, higher-intent stream is opening for the companies that get cited. The pipeline isn't only shrinking. For some companies, it's moving — away from those who optimized for the click and toward those who earn the citation.

A fair caveat: not every query is affected equally

One honest qualification, because the doomsday version of this story overstates it and you'll lose credibility internally if you do too.

AI Overviews show up most on informational queries — "how does X work," "what is Y," "best way to Z" — and far less on commercial and transactional ones, where the buyer is closer to a decision and Google has a stronger incentive to send the click (and the ad). If your traffic is concentrated on bottom-of-funnel, high-intent pages — pricing, comparisons, "near me," "vendor for X" — the click erosion is real but more contained than the headline numbers suggest. It's worth actually segmenting your own search console data by query intent before you panic or relax; the average hides the distribution.

But don't take too much comfort there, because of how funnels work over time. The top of the funnel is where you build the awareness and trust that the bottom of the funnel eventually converts. If AI Overviews quietly absorb your informational content — the articles that introduce people to your thinking, that earn the first unit of trust — you can keep your transactional clicks for a while and still watch the pipeline dry up a year later, when the awareness layer you stopped feeding finally runs out. The erosion at the top shows up as a shortage at the bottom, on a delay. That lag is precisely why most leaders won't connect the two until it's late: the cause and the symptom are a year apart, in different metrics, owned by different people.

What to do about it

The strategic response isn't to fight the AI Overview. It's to change what you're optimizing for, and to start before the lag catches up with you. Four moves:

1. Measure citation, not just ranking. Your dashboard almost certainly tracks position and impressions. It almost certainly does not track whether you're being cited in AI Overviews, or named in ChatGPT, Perplexity, and Claude answers, or how much referral traffic those tools send. Start watching all three. Citation is the metric that now predicts visibility, and you cannot manage — or defend the budget for — what you don't measure. This single change in what you report upward reframes the whole conversation from "why is traffic down" to "are we becoming the answer."

2. Rebuild your most important pages to be quoted. Take the pages that matter most to the business and restructure them around the citation patterns: a clean, self-contained answer near the top; genuine question-and-answer structure; visible expertise and cited sources; structured data. Not more words — more liftable words. A page an AI can quote cleanly is a page that wins, and most of your existing pages were built for a crawler, not a model. This is a finite, prioritizable project, not a boil-the-ocean rewrite: start with the ten pages tied to the most revenue.

3. Treat AI referral traffic as a channel worth building deliberately. It's small now and it converts at a premium. Track which AI tools send you visitors, what they land on, and what they do once they arrive. Then feed the machine that feeds you: the content structures that earn citations (clear answers, evidence, structure) are the same ones that compound this channel. The companies building citation authority in 2026 are buying a position in a channel that's only going to grow — at today's prices.

4. Segment before you strategize. Pull your own search console data and split it by query intent. Find out exactly which of your pages are losing clicks to AI Overviews and which are insulated. The right response for an informational page (restructure for citation, accept the click loss, harvest the referral) is different from the right response for a transactional one (defend the click, sharpen the offer). A blanket reaction wastes effort on pages that don't need it and under-invests in the ones that do.

The leaders who navigate this well won't be the ones who ranked the hardest. They'll be the ones who understood, early, that the job changed — from earning the click to being the answer — and who rebuilt their content and their measurement around that before the pipeline forced the conversation. Your rankings can stay exactly where they are while your traffic disappears. The only way through is to stop optimizing for a click that's no longer coming, and start earning the citation that is.

If you want to see where your own content stands against this shift — which pages are quietly going zero-click, and what it would take to become the cited answer instead — that's exactly the kind of question our architecture engagement is built to answer.

Frequently Asked Questions

What is an AI Overview and why does it reduce clicks?

An AI Overview is the AI-generated summary Google places at the top of search results, answering the query before any links. It reduces clicks because users get their answer without visiting a site. Pew Research found that when an AI summary appears, website clicks drop from 15% of searches to 8%, and only 1% of users click a link within the summary.

Why is my organic traffic dropping even though my rankings are stable?

Because ranking and traffic have decoupled. A page can hold position one and still lose most of its clicks if an AI Overview answers the query above it. Ahrefs measured position-one click-through rate falling roughly 78% when an AI Overview is present. Your rankings can look healthy while your traffic quietly collapses — which is why dashboards that track only position and impressions miss the problem.

Is SEO dead because of AI Overviews?

No, but the goal has changed. Ranking still influences whether you're cited — BrightEdge found a majority of AI Overview citations come from organically-ranked pages. The shift is that the conversion event moved from the click to the citation. The new objective is to be the source the AI quotes, not just to appear in the link list beneath it.

How do I get my content cited in AI Overviews and ChatGPT?

Structure content to be easily quoted. Semrush's large-scale study found the strongest predictors of AI citation were clear, summarizable writing (+33%), demonstrated expertise and trust signals (+31%), question-and-answer structure (+25%), clean section structure (+23%), and structured data (+22%). In short: well-organized, trustworthy, liftable content that itself cites credible sources.

Does Google AI Overviews affect all searches equally?

No. AI Overviews appear most on informational queries ("what is," "how does") and less on commercial or transactional ones where buyers are closer to a decision. If your traffic skews bottom-of-funnel, near-term erosion is more contained — but the top-of-funnel awareness loss tends to show up as a pipeline shortage later, so segment your own data before deciding how to respond.

Is AI referral traffic from ChatGPT and Perplexity worth anything?

Yes — disproportionately. It's small in volume today (under 1% of referral traffic by BrightEdge's measure) but high in quality. HockeyStack found LLM-sourced visitors converted to hand-raisers at a median 5.24%, with 86% classified as high-intent. People who arrive because an AI recommended you tend to arrive pre-qualified. ChatGPT drives the large majority of that referral traffic today, with Perplexity and Gemini secondary.

What's the single most important thing to start tracking?

Citation. Most analytics dashboards track ranking and impressions but not whether you're being cited in AI Overviews or named in AI assistant answers, or how much traffic those tools refer. That citation signal now predicts visibility better than ranking alone — and you can't improve, or fund, what you aren't measuring.

How is "answer engine optimization" different from SEO?

Traditional SEO optimizes to rank in the link list and earn the click. Answer engine optimization (sometimes called generative engine optimization) optimizes to be quoted in the AI's answer. The tactics overlap — quality, structure, authority all help both — but the target differs: a self-contained answer near the top, Q&A structure, cited evidence, and structured data matter more when the goal is being lifted into a summary than when it's ranking a page.

Sources

• Pew Research Center — Google users are less likely to click on links when an AI summary appears (July 22, 2025) — 8% vs 15% click rate, 1% in-summary, 26% vs 16% session end
• Ahrefs — AI Overviews Reduce Clicks update (December 2025) — position-1 CTR ~1.6% with AIO vs 7.3% prior; 3.9% without
• Seer Interactive / Search Engine Land — AI Overviews drive drop in organic CTR (November 4, 2025) — 1.76% → 0.61% informational CTR; +35% clicks (and +91% paid) if cited
• Google Blog — AI in Search is driving more queries and higher quality clicks, Liz Reid (August 6, 2025) — the counter-claim
• Google Blog — AI Mode in Search (May 20, 2025) — rollout + ~1B users + ~3× query length
• BrightEdge — One Year Into Google AI Overviews (May 14, 2025) — impressions +49% YoY, CTR ~−30%
• BrightEdge — AI Overview citation / rank-overlap analysis (September 2025) — majority of citations from organically-ranked pages
• Semrush — Content Optimization for AI Search study (January 14, 2026) — citation predictors (clarity +33%, EEAT +31%, Q&A +25%, structure +23%, structured data +22%)
• HockeyStack Labs — LLM Traffic in 2025 (May 30, 2026) — 5.24% median conversion, 86% high-intent; channel mix ChatGPT ~82% / Perplexity ~12% / Gemini ~5%
• bosio.digital — internal Google Search Console + behavioral analytics (June 1–7, 2026) — first-party zero-click pages (pos 6.0 / 887 imp / 0 clicks; pos 5.77 / 0 clicks)