SEO Pulse: Where Clicks Go, What Agents Skip, Who’s Leaving Bing

This week in SEO: a controlled test challenged Google’s explanation for lost clicks, a case study revealed why many Core Web Vitals fixes aim at the wrong target, new research measured how AI agents handle pricing pages, John Mueller offered guidance on agent access, and a longtime Bing leader announced his departure.

Here is what matters for your strategy.

AI Overviews Don’t Just Absorb Low-Value Clicks

New data from a randomized field experiment on Google’s AI Overviews found no measurable difference in bounce rates, return to search, or time on site between clicks that occurred with summaries and those that occurred without them.

The study measured a 39.8% drop in organic clicks when AI Overviews appear. Losses concentrated on informational queries, while navigational and transactional queries showed no measurable change on smaller samples. Google VP of Search Liz Reid has said AI Overviews cut “bounce clicks,” the low-value visits users abandon quickly, but has not released data to support that.

Why This Matters

If AI Overviews were mainly absorbing low-value visits, the extra clicks websites get when the summaries are removed should look worse, but they did not. The added clicks carried the same bounce rates, dwell time, and return-to-search behavior as the rest.

That leaves Google’s click-quality defense unsupported by the experiment’s data. A drop in clicks on AI Overview queries cannot be waved off as just losing visitors who would not have converted. Reid has repeated the bounce-clicks explanation in several public settings, and Sundar Pichai addressed the same traffic question in May, but Google still has not released the segmented click data that would settle it.

LCP Fixes Often Target The Wrong Element

Google’s John Mueller pointed to a case study explaining why so many Largest Contentful Paint (LCP) fixes fail to improve the score. In store layouts that vary by merchant, the browser can lock onto the wrong element, so every optimization after that targets something that was never the LCP.

The case study, published on web.dev, traces a year of Core Web Vitals work. The team traced the store’s weak LCP scores to the browser locking onto the wrong on-page element, a side effect of how its templates loaded, then adjusted the pages so the browser measured the real main content. The retailer reports that a higher share of its online storefronts passed LCP afterward.

Why This Matters

The useful part here is the order of operations, more than the specific fixes. Before compressing another hero image, confirm which element the browser actually counts as the LCP, because in template-driven or carousel-heavy layouts it may not be the one you think it is. HTTP Archive data shows the same pattern, where real-world LCP tends to break when the platform is slow to let the browser discover the main image, not only when a page is heavy.

What People Are Saying

The post drew a busy thread on Mueller’s LinkedIn, with more than 60 comments, and most of them centered on whether faster pages actually drive conversions.

The doubts came in two camps. On causation, Manhal Abou Zaki, an SEO manager at Omnicom Media Group, called the link indirect. He said a faster-loading page can certainly support and facilitate conversions, but it is unlikely to convince someone to convert solely because the page loaded more quickly. Francisco Antonio Fuentes Figueroa, an SEO consultant, went after the reported conversion lift itself, noting it was a full-year pre/post comparison with other changes shipping alongside it and no visible control group, and asking whether any A/B or staggered-rollout data existed to isolate the LCP fix. David Swinstead, who posts as The CRO Standup Comic, said he advocates for the speed-conversion link but still balked at its size. The counterweight came from Georgi Petrov, founder of Uxify, who said his own 50/50 tests run the other way, that LCP changes move conversion more than people expect.

The second camp went at rankings. Vijay V., a head of SEO, asked whether Core Web Vitals are a ranking factor at all, saying he has not seen any evidence they are. Mueller’s reply compressed the debate into a line. Vishal Parmar backed the doubt with field observation, noting sites with poor LCP, CLS, and INP that rank fine for competitive terms.

Others took the broader point the case study is making. Waqar Abdullah said the real lesson is to treat the LCP element as business-critical, not only to make the site faster. Amarachi Kalu put it as revenue over checkbox, that conversions are what businesses remember. Mat Bennett kept it light, noting the only place he still meets layout shift is in Google products.

AI Agents Fall Back To Third-Party Sources When Pages Won’t Load

A report from Siteline ran a simulated Claude agent against 100 top B2B software products to see whether it could find their pricing. When the agent hit access errors or pricing it could not read, it often pulled the numbers from third-party sites instead.

Siteline founder David Kaufman had the agent try to report prices and features across a set of top B2B software sites. On a notable share of attempts, the agent hit an access error or unreadable pricing, and in those cases it was far more likely to abandon the brand’s own page and pull numbers from outside sources that can be stale. Most of the failures traced back to pricing loaded with JavaScript that the agents do not render, or to prices hidden behind a sales contact.

Why This Matters

An AI agent is now a visitor you can accidentally lock out. A page that looks complete in a browser can read as empty to an agent if the important parts load client-side after the initial fetch. When that happens, the agent does not stop. It finds the answer somewhere else, often on a third-party site or a competitor that lists its prices in plain text. The fix is to make pricing and key features readable in the initial HTML rather than loading them afterward, and surface them prominently on the page.

Mueller Says Don’t Blindly Block Agentic Browsers

John Mueller was asked whether Google’s quality principles will change as AI agents browse sites for users. His answer was that most of the principles hold, with one new best practice worth noting.

The question, put to Mueller on Bluesky, was whether guidance like “images provide a satisfying experience” still applies when the visitor is an information agent rather than a person. Mueller said he expects most principles to stay the same, because a site that is useful to people is generally useful to agentic browsers too. He added that some details will evolve, and named not blindly blocking agentic browsers as a new best practice that will come into play.

Why This Matters

Mueller’s answer draws a line between two different things. Content quality standards are not being rewritten for the agentic era, because those standards were written for humans and agents are serving those same humans. What is new is technical accessibility. A site can meet every quality bar and still hurt itself by blocking the agents that increasingly act on a user’s behalf. This mirrors the early days of nofollow, when some sites sculpted their internal links to hoard PageRank and ended up starving important pages. Blocking agentic browsers could follow the same pattern, a technical decision made for one reason with unintended visibility costs later. The takeaway here is to check your bot and access rules before they quietly cut off a class of visitor Mueller says not to block blindly.

Bing’s Fabrice Canel Announces Retirement

Fabrice Canel, the Principal Product Manager who led Bing’s crawling and indexing team and championed IndexNow, announced his retirement from Microsoft after nearly 30 years. He shared the news in a LinkedIn farewell post, writing that he had accepted Microsoft’s Voluntary Retirement Program, effective July 1.

His exit removes one of Bing’s most recognizable contacts for the SEO and webmaster community, at a point when Bing’s index quietly powers AI search products like Copilot and ChatGPT’s web results. That role is why Bing’s index still matters, even as its consumer search share remains small.

Theme Of The Week: The Plumbing Under Search

The AI Overviews experiment is about measurement, whether the traffic Google takes was ever worth much. The LCP case study is a measurement story too, to see whether the browser is scoring the element you think it is. The Siteline report is about access, whether an agent can read your page well enough to send a buyer to your site. Mueller’s answer is about the rules of that access, and where blocking agents turns into a visibility problem. Canel’s retirement is the human version of the same story, one of the people who built and explained that layer stepping away. Put together, the week is less about a new feature and more about the machinery underneath, how sites get measured, read, and reached as more of the audience arrives through an AI layer. The work of staying visible is moving one level down, into whether the systems in front of your readers can use what you publish.