Zero-click search: why fewer clicks make AI visibility tracking non-negotiable

· Prompt Ledger editorial team

The short version

A zero-click search is one where the searcher gets their answer on the results page itself and never clicks through to a website. In the US, 68% of Google searches were zero-click in early 2026, up from 49% in 2019. When an AI Overview appears on the page, organic clicks drop by roughly 40% and the zero-click rate on that query jumps to around 68% from a 49% baseline without one. Google's AI Mode runs 92-94% zero-click. None of that traffic shows up in a rank tracker - which is why mention and citation rate inside the answer itself are becoming the metric that matters.

"Zero-click search" is an old SEO term that has quietly picked up new urgency. It used to mean featured snippets and knowledge panels answering simple questions directly on the results page. It now mostly means something bigger: a generated AI answer - an AI Overview, AI Mode, Perplexity's synthesis, ChatGPT's response - sits between the searcher and the list of links, and increasingly the searcher never gets past it.

How the zero-click rate crept from 49% to 68%

The trend predates generative AI by years, but it has accelerated sharply since AI-generated answers started appearing on the results page. Jumpshot's older clickstream data put the US zero-click rate at 45% in 2016 and 49% in 2019. Semrush and Datos measured it at roughly 60% in 2024. By early 2026, a Similarweb analysis circulated by SparkToro put it at 68% in the US, with the UK slightly higher at 69.5%.

US zero-click search rate over time
PeriodZero-click rateSource
2016~45%Jumpshot clickstream data
2019~49%Jumpshot clickstream data
2024~60%Semrush / Datos
Early 2026~68%Similarweb analysis, via SparkToro

The pace of change is the more interesting number. The rate moved about 2.3 points a year between 2019 and 2024; from 2024 to early 2026 it moved closer to 3.8 points a year. That acceleration lines up with when AI Overviews rolled out broadly, not with any change to classic featured snippets.

What an AI Overview does to a single query's clicks

The aggregate trend is the headline, but the per-query effect is sharper and easier to act on. A randomized study by Agarwal and Sen, alongside a separate Ahrefs analysis of roughly 300,000 keywords, found that when an AI Overview appears on a search results page, organic clicks on that page drop by about 40% and the query's zero-click rate jumps to around 68% from a 49% baseline without an AI Overview present. The position-1 organic result takes the worst of it: about a 58% drop in click-through rate on queries that trigger an AI Overview, compared with the same position on a query that does not.

Mobile searches are already closer to that no-click world than desktop: roughly 77% of mobile US Google searches end without a click versus about 51% on desktop, which tracks with how much more screen real estate an AI-generated answer or snippet takes up on a small screen before any blue link appears.

Google AI Mode is the extreme case

AI Mode - Google's full conversational search experience - reportedly runs a 92-94% zero-click rate, per SparkToro and Semrush tracking. It is still a small share of total Google queries, under half a percent by most estimates, but it is the clearest preview of where the regular results page is heading: an answer written for you, with the underlying sources reduced to small citation chips most people never tap.

Why rank and click-through rate stop being the full scorecard

A rank tracker tells you where a URL sits in a list of blue links. That number was always a proxy for something else - whether people found and used your page - and the proxy has started to break down. On a query that triggers an AI Overview, you can hold position one and still lose the majority of the clicks that position used to deliver, because the searcher's question got answered above you. Position itself did not get worse; what position buys you did.

This is the same shift AI search optimization and answer engine optimization are both built around, from a different angle: instead of starting from the question "how do I rank higher," zero-click data starts from the question "what does ranking even deliver anymore on this query."

What to track once the click is not the signal

If a meaningful share of your audience never reaches your site for a given query, clicks and sessions cannot be the only thing you measure. Three numbers fill the gap, and they are the same three a GEO or AI visibility tool is built to sample on a schedule:

What replaces the click on a zero-click query
MetricWhat it tells youWhere it shows up
Mention rateDoes the AI answer name your brand at all, for questions your buyers ask?AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini answers
Citation rateWhen mentioned, does the answer link to one of your pages?Visible source chips in AI Overviews and Perplexity; footnotes in ChatGPT and Gemini
Share of voiceWho gets named alongside you, and how often do they take the spot you wanted?Competitor names appearing across the same sampled questions

None of those three show up in Google Search Console, because Search Console only logs an impression and a click on a result that gets rendered as a link - not whether an AI-generated answer quoted your page without a click ever happening. That blind spot is exactly why a dedicated tracker, sampling the same questions repeatedly across engines, has become a separate line item rather than a feature of a classic rank tracker.

You do not need a paid subscription to get a first read. Our GEO audit guide walks through running that same check by hand - asking the questions your buyers ask across a few assistants, more than once, and logging whether you were named and cited. It is slower than a dashboard and it does not scale, but it tells you in an afternoon whether this is a problem worth paying to monitor.

Once it is worth monitoring on a schedule, that is the job our ranking of AI visibility tools covers, and the individual reviews go deeper on what each one actually samples and how often.

Questions

What is a zero-click search?

A search where the person gets their answer directly on the results page - from a featured snippet, knowledge panel, or increasingly an AI-generated summary - and never clicks through to a website. In the US, roughly 68% of Google searches were zero-click in early 2026.

Has zero-click search gotten worse because of AI Overviews?

The rate was already rising before generative AI - from about 45% in 2016 to 49% in 2019 to roughly 60% in 2024 - but it accelerated once AI Overviews rolled out broadly. A randomized study and a separate Ahrefs analysis of 300,000 keywords both found organic clicks drop by about 40% on queries where an AI Overview appears, with the zero-click rate on those queries jumping to around 68% from a 49% baseline.

Does ranking #1 still matter if the search is zero-click?

It still matters, but less than it did. Position 1 on a query that triggers an AI Overview sees roughly a 58% drop in click-through rate compared with the same position on a query without one. The page can hold its rank and still lose most of the clicks that rank used to be worth.

How zero-click is Google's AI Mode specifically?

Reported at 92-94% zero-click, per SparkToro and Semrush tracking - substantially higher than the roughly 68% rate for Google search overall. It currently makes up a small share of total queries, but it shows the direction regular search results are moving in.

If people aren't clicking, how do I know whether my content is working?

Track mention rate (does the AI answer name your brand), citation rate (does it link to your page when it does), and share of voice (who else gets named on the same questions) instead of relying on clicks alone. These require sampling the same questions repeatedly across assistants, since answers vary between identical prompts - which is what a GEO audit or a dedicated AI visibility tool does.

Can I measure this without buying a tool?

Yes, as a manual starting point. Our GEO audit guide sets out how to ask the same buyer questions across a few assistants by hand and log whether you were named and cited. It does not scale the way a scheduled tracker does, but it is enough to tell you whether the problem is worth monitoring.

Mentioned in this note

Back to the Best Generative Engine Optimization Tools results