Field Notes · GEO
Zero-Click Search: Visibility Wins
The open web is being rationed, and the asset that compounds is no longer your session count but whether the answer engine repeats your name.
For two decades, search success had one scoreboard: sessions. You ranked, people clicked, your analytics filled up, and everyone agreed you were winning. That scoreboard is now quietly broken. The search result page increasingly answers the question itself, names a few sources, and sends the user on their way without a visit to anyone.
This is zero-click search, and it has crossed a threshold that changes how a serious marketing team should think. The point of this Field Note is not to alarm you about lost traffic. It is to argue that the click was always a proxy, and that the real prize has shifted upstream: being the brand the search engine, and now the answer engine, chooses to repeat.
What does zero-click search actually mean?
A zero-click search is one where the user gets what they came for on the results page itself and never clicks through to a website. The answer is delivered by a featured snippet, a knowledge panel, a People Also Ask box, a local pack, or, increasingly, an AI Overview that synthesises a full response at the top of the page.
Crucially, the click never disappears because the demand disappears. It disappears because the platform satisfied the demand in place. From the user's point of view this is a better experience. From a publisher's point of view it means the same search interest now produces a fraction of the visits it once did.
How big is the trend, and what does the best data show?
Zero-click is now the majority outcome of search by a wide margin, and it stepped up sharply over the last two years. According to SparkToro's 2026 study, built on Similarweb clickstream data, 68.01% of US Google searches ended without a click to the open web in the first four months of 2026, up from 60.45% in 2024. That is a 7.56-point rise in just two years.
Put in human terms: for every 1,000 US Google searches, only 374 clicks now reach the open web, and in the EU the figure is 360. The share of searches generating at least one click fell 9.51 percentage points between 2024 and 2026, a relative decline of 22.9%. Similarweb, reporting separately in July 2025, found zero-click Google searches rose from 56% to 69% in roughly a year, an inflection that lines up with the rollout of AI Overviews. Two independent measurement houses arriving at much the same picture is about as close to consensus as this field offers.
Why is this happening now?
It is happening because the search interface stopped being a list of links and became an answer surface, and AI accelerated a trend that was already well underway. This is the single most misunderstood part of the story, so it is worth correcting plainly: zero-click is not new and was not caused entirely by AI.
In the 2024 SparkToro and Rand Fishkin study, drawing on Datos clickstream data, 58.5% of US Google searches already ended in zero clicks, up from around 65% in 2021 by an alternative measure, long before AI Overviews existed. Featured snippets, knowledge panels, People Also Ask, and simply abandoned or refined queries were doing the work. What AI Overviews changed is the slope of the curve.
- AI Overviews now appear on more than 20% of Google searches, according to SparkToro's 2026 study, moving from novelty to a routine feature in under two years.
- When an AI Overview appears, click-through rates drop by nearly 60%, per the same SparkToro data.
- Pew Research Center, tracking around 69,000 searches from 900 US adults, found that when an AI Overview appeared users clicked a traditional result only 8% of the time, versus 15% without one, a relative reduction of roughly 47%.
- Only 1% of searches with an AI Overview produced a click on a link inside the summary itself, Pew found, and users ended their browsing session 26% of the time after seeing one.
Is traffic really collapsing, or is that overstated?
The magnitude of the traffic hit is genuinely disputed, and an honest strategy holds both views at once. On one side, clickstream studies and publishers describe steep declines. The UK Professional Publishers Association documented year-on-year CTR falls of 10% to 25% despite stable rankings, with one lifestyle publisher reportedly dropping from 5.1% to 0.6% CTR, though that is a single advocacy source and should be read as illustrative.
On the other side stands Google. In August 2025, Search head Liz Reid said total organic click volume to websites was relatively stable year-over-year and that average click quality had slightly increased, while John Mueller said clicks from AI Overview pages are higher quality. Both sides have incentives, and publisher internal data reportedly does not support the higher-quality claim. The defensible reading is this: the easy, low-intent informational click is being absorbed by the answer surface, while the high-intent click that signals real buying interest still exists. It is simply rarer, and therefore worth more.
What should you measure instead of sessions?
You should measure whether your brand is present in the answer, not just whether someone clicked through to your page. Traditional CTR and session counts can no longer tell you whether you were named inside a response the user read and acted on but never clicked. Visibility now has to be tracked at the answer and prompt level.
A new measurement category has emerged to do exactly this, often called AI Share of Voice: your brand mention rate, your positioning within answers, and your comparative share against rivals, tracked per engine across ChatGPT, Perplexity, and Gemini. AthenaHQ's State of AI Search 2026 report put the average brand mention rate across AI answers at 17.2%, a figure worth treating as a vendor benchmark rather than settled industry fact. The engines also behave differently: a 2026 citation-measurement framework analysing 21,143 citations found Perplexity cites more sources per prompt but with lower influence each, while ChatGPT cites fewer sources and extracts more from each one. That distinction should shape where you focus.
- Impressions and presence: how often you surface in answers and panels, not just sessions captured.
- Brand mention rate and AI share of voice: how often you are named, and where you sit relative to competitors, per engine.
- Citation rate: how often your specific pages are the attributed source inside an AI answer.
- Assisted conversions and branded search lift: the downstream signal that being seen, even without a click, is feeding demand.
How do you adapt without panicking?
You adapt by optimising to be cited and remembered rather than merely clicked, because being the named brand is now upstream of, and more durable than, the click. Inside an AI Overview just 1% of users click a cited link, so the citation itself, not the onward visit, is the asset. The work is mostly disciplined, not dramatic.
- Write answer-first. Make every heading a clear question or claim, and have the first sentence beneath it answer outright, so a retrieval system can lift a clean, quotable passage.
- Earn the citation. Publish specific, defensible facts, original data, and clear definitions that an answer engine can attribute with confidence, rather than thin restatements of consensus.
- Protect high-intent journeys. Accept that informational queries will resolve on the SERP, and concentrate conversion effort on the rarer, more valuable commercial searches that still click.
- Instrument AI share of voice now. You cannot manage what you do not measure, and brand mention rate across the major engines is the metric that reveals whether you are winning the answer.
One forward-looking caveat keeps this honest. Google AI Mode is still tiny in measured use, just 0.34% of searches in early 2026 per SparkToro, yet Google stated at I/O 2026 that it passed a billion monthly users with query volume more than doubling each quarter. Anything you read about its impact is a projection, not a fact. Build for the world where answers, not links, are the interface, and you are ready either way.
The scoreboard changed while the game looked the same. Traffic counts will keep falling for reasons that have nothing to do with how well you are doing, and the teams that thrive will be the ones measuring whether the answer engine repeats their name. That reframe, from sessions captured to answers owned, is precisely the work we do at DarkHorse: getting you found on Google, cited by AI, and built to convert when the rare high-intent click finally lands.
Questions people ask
Does zero-click search mean SEO is dead?
No, it means the goal of SEO has shifted from capturing clicks to earning visibility and citations. Ranking still matters because answer engines pull from highly ranked, well-structured pages to build their responses. The discipline is evolving from sessions-chasing into answer optimisation, not disappearing.
How much of search is now zero-click?
According to SparkToro's 2026 study using Similarweb data, 68.01% of US Google searches ended without a click to the open web in early 2026, up from 60.45% in 2024. Similarweb separately reported a rise from 56% to 69% over roughly a year. Both point to zero-click being the clear majority outcome of search today.
Did AI Overviews cause the zero-click trend?
AI Overviews accelerated an existing trend rather than starting it. Zero-click already sat near 58% to 65% before AI Overviews existed, driven by featured snippets, knowledge panels, and People Also Ask boxes. AI Overviews, now on more than 20% of searches, sharpened the slope by dropping click-through rates by nearly 60% when they appear, per SparkToro.
What should I measure if not traffic?
Measure brand presence at the answer level: your AI share of voice, brand mention rate, citation rate, and assisted conversions, alongside impressions. Sessions alone cannot tell you whether your brand was named inside an answer the user never clicked. Tools tracking mention rate across ChatGPT, Perplexity, and Gemini are becoming the North Star metric for this kind of visibility.