The Click You're Not Getting Anymore: How Businesses Actually Earn a Place Inside AI Search

Somewhere in the last two years, the Google results page quietly stopped being a list of ten blue links and started being a place where a lot of questions get answered without anyone going anywhere. If you've noticed your traffic doing something strange — impressions climbing while clicks stay flat, or a keyword you rank well for suddenly sending you fewer visitors than it did a year ago — you're not imagining it, and it's not a ranking problem in the way ranking problems used to work. Something upstream of your rank changed.

This isn't a piece about whether AI is "coming for search." It already arrived, it's already reshaping how many people never leave the results page at all, and the interesting question at this point isn't whether it's happening — it's what a business actually does about it, concretely, this quarter, that isn't just "wait and see."

The number that changed the job

Start with the plainest version of what's different. In the first four months of 2026, a whopping 68.01% of Google searches ended without a click. That figure comes from SparkToro's long-running tracking of clickstream data, and it's not a fringe estimate — it's up from 60.45% in 2024, a rise of roughly 7.5 percentage points in two years. Put plainly: a majority of the searches happening on Google right now resolve without a single visit to any website, yours or a competitor's.

That number by itself would be alarming enough. What makes it more specific — and more useful, if you're trying to actually do something about it rather than just worry — is what happens on the subset of searches where an AI Overview renders. Pew Research Center ran a behavioral study tracking 68,879 real Google searches, not survey responses, and found that when an AI Overview is present, users clicked a traditional result only 8% of the time — compared to 15% without one. That's not a small dip. That's roughly half the click behavior, measured by watching what people actually did rather than asking them to remember or guess.

None of this is static, either. Seer Interactive tracked 53 brands across 5.47 million queries and 2.43 billion impressions from January 2025 through February 2026 and found that after bottoming out at 1.3% in December 2025, the click-through rate on Google's AI Overviews climbed to 2.4% in February 2026 — an 85% jump in two months. That sounds like good news, and it partly is, but read the comparison line carefully: over the same window, CTR on queries without an AI Overview increased from 2.8% in early 2025 to 3.8% by February 2026. The gap isn't closing. Both lines are rising, and the AI-Overview line is still well behind. Google's own AI features, meanwhile, keep expanding — the company said at I/O 2026 that AI Mode had surpassed 1 billion monthly users and that query volume was more than doubling each quarter. This isn't a temporary artifact of a rough algorithm update. It's the shape of where search is heading.

Ranking and getting cited are no longer the same job

Here's the part that trips up a lot of otherwise sharp SEO strategy: ranking well and being the source an AI Overview quotes are related, but they are not the same achievement, and the gap between them is widening. One of the more striking data points to come out of recent AI Overview research is that only 37.9% of URLs cited in AI Overviews also rank in the top 10, down from around 76% in July 2025, based on an analysis of 4 million AI Overview URLs.

Sit with that for a second, because it cuts against the intuitive assumption that "just keep ranking #1 and you'll get cited automatically." A year ago, being in the top 10 was a fairly reliable predictor of being pulled into an AI-generated summary. That relationship has weakened substantially. Whatever Google's citation system is now weighing, it isn't simply reading down the existing ranked list and grabbing the top few. It's evaluating something closer to "which page states this fact most clearly, most verifiably, and with the least ambiguity" — which is a genuinely different question than "which page has historically earned the most authority signals for this keyword."

That distinction matters because it means a well-ranked page can lose ground to a less-authoritative page that simply states the answer in a cleaner, more extractable way. And it means the old SEO instinct — write comprehensively, cover every angle, hedge every claim — can actively work against you in the citation game, where a flatly stated, well-sourced fact beats a nuanced five-paragraph exploration of the topic.

Why the traffic that's left is worth more, not less

It would be easy to read the numbers above as pure decline. There's a second half of the story that's easy to miss if you stop at the click-through-rate charts. Being cited inside an AI Overview isn't nothing — it's measurably better than being on the same results page and not cited. The same Seer Interactive data found that when an AI Overview appears and cites a given page, that page earns roughly 2.1% CTR, while an uncited page on that same results page earns roughly 0.9% — both trailing the no-AI-Overview baseline of about 3.3%, but being cited clearly beats being ignored. Separate research from the same firm found the gap is even larger in absolute terms: brands cited inside AI Overviews earn 35% more organic clicks and 91% more paid clicks than non-cited brands on the same query.

There's also a quality difference in who actually clicks through. A person who reads an AI-generated summary and still chooses to click a source has, by definition, already decided the quick answer wasn't enough — they want more. That's a self-selected, higher-intent visitor, not someone skimming ten blue links looking for the fastest exit. This is the same logic that shows up anecdotally across e-commerce case studies: a small number of AI-referred visitors converting at a rate that dwarfs ordinary search traffic isn't a coincidence, it's what you'd expect when only the genuinely interested people are still bothering to click at all.

None of this means the traffic decline isn't real, or that businesses should feel relaxed about it. It means the right response isn't "panic about the click number" — it's "make sure you're one of the sources getting cited," because that's now the group capturing whatever attention is left.

Schema markup: useful, overhyped, not a shortcut

If you've spent any time reading about this shift, you've almost certainly run into confident claims that structured data — schema markup — is the key to getting cited by AI. Some of that is true. Much of it is overstated, and it's worth separating the two, because implementing schema badly and calling it done is a common and avoidable mistake.

What schema actually does is unglamorous: it labels facts on your page in a machine-readable format, so a system trying to extract "who wrote this, what does it claim, when was it published, who is the business behind it" doesn't have to guess. Google itself has been fairly direct that no special schema is required for AI Overviews or AI Mode — a page must still be indexed and eligible to appear with a search snippet, and that structured data should match the visible page. In other words: schema is not a magic switch. It doesn't override the fact that a page needs to be indexed, well-written, and genuinely relevant before it's eligible for anything.

An Ahrefs study tracking 1,885 pages that added schema puts real numbers behind that caution. The catch in the study is important: the researchers studied pages that were already being cited heavily by AI — every page in the dataset had 100+ AI Overview citations in February 2025, before any schema was added — meaning these pages were already inside the consideration set, being crawled and surfaced by LLMs. For that group, adding schema barely moved the needle. The honest interpretation isn't "schema doesn't matter." It's "schema doesn't rescue or elevate a page that's already winning" — but the same researchers noted that for pages that aren't being seen by AI systems at all, schema markup might still play a role in helping them get crawled, parsed, or indexed in the first place.

So the practical takeaway sits in the middle of the hype and the skepticism: implement schema properly — Organization markup with a consistent name and canonical @id, Article markup that correctly separates author from publisher, and FAQPage markup on pages structured as genuine question-and-answer content — because it removes ambiguity and helps a page get into consideration in the first place. Just don't expect it to substitute for the page actually being a clear, well-sourced answer to something real people are asking.

The trust framework AI can't fully fake its way around

Underneath all of this — the citations, the schema, the ranking shifts — sits a framework Google has been refining for over a decade and that matters more, not less, in an AI-mediated results page: E-E-A-T, short for Experience, Expertise, Authoritativeness, and Trustworthiness. It's a framework defined in Google's Search Quality Rater Guidelines — a 176-page document that Google gives to thousands of human quality raters who manually evaluate search results. It's worth being precise that E-E-A-T is not itself a ranking algorithm input; it's the lens human raters use to judge whether Google's actual ranking systems are surfacing good content, and that feedback trains the systems at scale.

The framework got its second "E" — Experience — in a meaningful update. Google first introduced the E-A-T concept in its Quality Rater Guidelines in 2014, and it was the gold standard for content quality assessment for years, until December 2022, when Google added a second E for Experience, recognizing that first-hand, real-world experience with a topic is a distinct and valuable quality signal, different from academic expertise. That addition wasn't cosmetic. It was Google explicitly acknowledging that someone who actually built the deck, actually lived through the renovation, actually ran the campaign, brings something a well-researched but secondhand article can't replicate.

Of the four components, trust is the one Google has been most explicit about weighting above the rest. One widely cited line from the guidelines states plainly that trust is "the most important member of the E-E-A-T family" — the reasoning being that experience, expertise, and authority all lose their value if the underlying content can't be trusted to be accurate. Practically, that shows up as things like comprehensive author attribution with verifiable credentials, transparent contact information and business verification, and security infrastructure including HTTPS as a baseline expectation. None of that is exotic. Most of it is the kind of thing a rushed website build skips because it doesn't show up in a screenshot — a real author bio instead of "Admin," a real phone number instead of a contact form that goes nowhere, a physical address that matches what's on Google Business Profile.

This matters more, not less, in an AI-mediated world, for an almost paradoxical reason: when a machine is deciding which claim to lift and attribute, it's leaning on exactly these same signals to decide who's a credible enough source to cite without embarrassing itself. An AI system extracting a fact still has to decide whose fact to trust, and the framework it's implicitly trained against is the same one built for human quality raters over the last decade.

What actually earns a citation

Put the last two sections together and a fairly concrete picture emerges of what separates a page that gets pulled into an AI Overview from one that doesn't, beyond simply "being good."

State the answer plainly, early, before the hedging. AI extraction systems are looking for a clean, quotable, verifiable claim. A paragraph that builds up context for four sentences before finally answering the question buries the exact sentence a machine would want to lift. This is uncomfortable for anyone trained to write nuanced, caveat-heavy content, but it's the format that gets extracted.

Structure content as actual question-and-answer where that's genuinely the format. FAQPage markup and FAQ-formatted sections consistently show up across the research as one of the highest-leverage structures, precisely because it explicitly maps content to the question-answer format AI engines extract. That doesn't mean bolt an FAQ onto every page for the sake of it — it means recognize when a page's content is naturally answering discrete questions, and format it that way instead of forcing it into narrative prose.

Make the author a real, checkable person. A named author with a real bio, real credentials, and a real track record does more for citation eligibility than almost any technical tweak, because it's solving the exact trust problem described above — a machine deciding whether to attribute a claim to you is running the same evaluation a skeptical human would.

Keep facts internally consistent, everywhere. A price, a date, a claim that appears in your visible text needs to match what's in your schema, your other pages, and your Google Business Profile exactly. Mismatches don't just look sloppy to a human — they read as a trust failure to a system trying to decide if your page is reliable enough to quote.

Get indexed and crawlable before anything else. This sounds too basic to mention, but the Ahrefs research above makes the point directly: a page with beautiful schema and perfect E-E-A-T signals that isn't indexed can't be cited, full stop. Indexing is the gate everything else sits behind.

The local business version of this problem

If you run a local service business, there's a temptation to treat all of this as a "big content site" problem — something that matters for publishers and SaaS companies chasing informational keywords, not for a roofer, a dentist, or a boutique law firm competing on "near me" searches. That assumption doesn't hold up against the data.

Seer Interactive's research breaking down AI Overview trigger rates by query type found something specifically relevant here: informational queries containing "near me" showed AI Overviews 76.9% of the time — and because local-intent searches are typically associated with maps and local packs, the fact that nearly 4 in 5 of "near me" informational queries are showing AI Overviews suggests Google is layering AI Overviews on top of local SERPs. That's a meaningfully different landscape than the one most local SEO advice was written for. It's no longer enough to win the map pack and the top three organic spots — there's now a third layer, an AI-generated summary, potentially sitting above both, drawing from sources that may or may not be your listing.

The practical implication is that the same E-E-A-T fundamentals apply locally, just with local-specific evidence: a real named practitioner instead of an anonymous "our team," genuine reviews with specifics rather than generic five-star noise, an about page that actually describes who's behind the business rather than stock "family owned since" copy nobody can verify. The businesses most exposed here are the ones who treated their Google Business Profile as the whole local SEO strategy and never built out a website with enough real, structured, citable substance to be a source an AI system would want to pull from.

What this doesn't mean

It's worth being clear about what this shift is not, because the hype cycle around "AI is killing SEO" or, in the opposite direction, "just add schema and you'll dominate AI search" both oversell simple stories that don't match the actual data.

It doesn't mean traditional SEO fundamentals stopped mattering. Every mechanism described above — crawlability, indexing, backlinks, topical authority — still sits underneath the citation layer. AI Overviews and AI Mode are built on top of Google's existing search index, not a separate system running on different rules.

It doesn't mean every business needs to obsess over AI citation tracking starting tomorrow. If your business runs almost entirely on direct traffic, referrals, or a Google Business Profile that already converts well, the marginal value of chasing AI Overview citations is lower than it is for a business that depends heavily on informational, top-of-funnel search traffic. Know which kind of business you are before reallocating effort.

And it doesn't mean schema markup or an FAQ section is a silver bullet. The research above is fairly blunt that structured data helps at the margins — mostly by making an otherwise-invisible page visible — and does very little for pages that are already strong. Chasing technical fixes while ignoring the actual substance of the content is the same mistake it's always been, just with a new acronym attached to it.

Where this actually leaves you

The uncomfortable version of this article is short: fewer people are clicking through from Google than at any point in the platform's history, and the trend line points further in that direction, not back. The useful version is longer, and it's this — the businesses that show up inside the summary, rather than getting skipped past, are the ones treating trust, authorship, and clear, verifiable claims as the actual product, not decoration bolted onto content built for a different era of search.

None of the mechanics here are exotic. A real author with a real bio. Facts that match everywhere they appear. Content that answers the actual question in the first sentence instead of the fifth. Schema implemented correctly, understood for what it is — a clarity layer, not a growth hack. None of it requires reinventing how a website gets built. It requires treating the site as something a skeptical machine has to be convinced by, on top of everything it already had to convince a skeptical human of.