Google's AI Overview links are multiplying, but many still don't reach your site

Google is adding more links and previews to AI Overviews, but citation rates lag far behind. Here's why link volume and referral traffic are not the same thing.

Google has added more links, hover previews and clearer source icons to AI Overviews and AI Mode over the past year. But new analysis shows that having more links on screen does not mean more of them actually lead users to the original web pages. For founders and marketers, the takeaway is blunt: link volume and real referral traffic are two different metrics, and only one of them pays your bills.

What changed?

Google has been iterating on how links appear inside AI-generated answers. The most recent update came from Robby Stein, Google's VP of Product for Search, who said groups of links would now appear in a pop-up when a user hovers over them on desktop, alongside more descriptive link icons across devices, according to Search Engine Journal.

That builds on an earlier commitment from August 2025, when Stein described showing more inline links as the company's long-term goal for AI Mode, adding embedded link carousels on desktop at the time. He said fluctuation was expected over the following weeks as the rollout continued, with the stated aim being to show more inline links over the long term.

The bigger context is scale. AI Overviews now sit on top of a majority of US searches. According to Advanced Web Ranking data reported by Xponent21, Google AI Overviews now appear in 60.32% of US queries, the first time the feature has cleared the 60% mark, doubling its presence since August 2024. That same data points to ChatGPT closing in on a similar scale, with roughly 2.5 billion queries a day and a trajectory that could see it overtake Google's search volume by 2027, according to AirOps CEO Alex Halliday, cited in the same report.

Why does this matter for your business?

More links on screen sounds like good news for publishers and brands worried about AI answers replacing clicks. But the gap between links appearing and links actually being followed is exactly where AI visibility strategy lives or dies right now.

Separate research on citation rates backs this up. Even as AI Overviews expanded from 15% to 43% of Google searches in twelve months, and Google AI Mode visits rose from 126 million to 279 million, ChatGPT's citation rate, the mechanism that actually names and links a brand inside an answer, sits at just 6.8% of queries, even after growing fivefold.

In other words, a redesigned pop-up or a more prominent icon does nothing for a brand that was never cited in the answer to begin with. The visible design layer that Google keeps polishing sits on top of a much harder problem: whether your content gets pulled into the answer at all.

There is also a structural reason why more links do not automatically mean more traffic. Google's own product team has acknowledged the traffic concern is real and ongoing, with Search Engine Journal describing the work as one of several forces pulling attention, clicks and visibility away from the organic channel as practitioners have known it.

The same underlying issue shows up on the AI crawler side. Research from TollBit, cited in GPO's September state-of-search roundup, found that roughly 15% of AI page fetches from ChatGPT's bot on tracked European sites reached pages that were supposed to be off-limits, while Anthropic states all three of its bots consistently honour robots.txt. If crawlers behave unevenly on the way in, it should not surprise anyone that links behave unevenly on the way out.

What should you do now?

First, stop treating "we appeared in an AI Overview" as the finish line. The pop-up redesign matters for user experience, but it says nothing about whether your specific page was one of the sources cited. Track citations and referral traffic separately, not as a single combined "AI visibility" number.

Second, watch your server logs and analytics for actual AI-referred sessions rather than assuming design changes translate into clicks. Shopify's own data, also cited in GPO's roundup, shows AI-referred storefront sessions grew 197% year-on-year and converted at roughly twice the rate of organic in research-heavy categories, though organic search still drove more total traffic than all AI platforms combined. That is a reason to keep investing in AEO, not a reason to expect it to replace organic search volume any time soon.

Third, make sure your content is structured so that if you are picked as a source, the hover preview and link icon actually represent you well: clear titles, accurate meta descriptions and a recognisable brand name near the top of the page. None of that guarantees a citation, but it removes an easy reason to be skipped.

Finally, if you have not checked which AI systems are actually citing your brand versus simply mentioning your category, that is worth doing before investing further in any single platform's link design changes. A free audit, such as the one Sited runs at sited.online, can show you where your citations currently stand across the major AI search engines before you decide where to put your next quarter of content effort.

Frequently asked questions

Does Google's hover pop-up link update mean more traffic for my site?

Not necessarily. The update changes how links are displayed, giving users a preview card before they click, but it does not change whether your page was selected as a cited source in the first place. Citation selection and link display are separate problems.

How common are AI Overviews now?

Google AI Overviews now appear in 60.32% of US queries, according to Advanced Web Ranking data reported by Xponent21, roughly double the share seen in August 2024.

Is ChatGPT catching up to Google Overviews in citation behaviour?

Not yet at the same rate. Even after strong growth, ChatGPT's citation rate sits at just 6.8% of queries, up fivefold over the past year, according to data reported by Something Inc.

Should I stop worrying about traditional SEO because of AI Overviews?

No. AI Overviews still rely heavily on the same web content that ranks in traditional search, so technical health, topical relevance and trustworthy sourcing remain foundational, even as the visible competition shifts toward citable passages rather than blue-link positions.

What is the single most useful thing to track right now?

Separate your citation rate, how often you are named and linked inside AI answers, from your referral traffic, how many of those citations turn into actual visits. Treating them as one number hides exactly the gap this story is about.

Sources