Ahrefs Data: AI Overviews Cut CTR by 23.1% After French Launch

The week's most concrete data point on AI Overview impact comes from Ahrefs, which tracked 963 domains via Google Search Console over 28 days before and 9 days after AI Overviews launched in France on July 22, 2026. The most exposed domains lost 23.1% of their click-through rate within just nine days of rollout — one of the largest and most rigorous measurements of AI Overview CTR impact published to date.

Meanwhile, Search Engine Land reports that Google Ads data shows a measurable shift in search query length following the AI Mode rollout, suggesting users are adapting their search behavior — a finding with direct implications for keyword strategy and content targeting.

Google is running a small test called the AI Contribution Pilot, which compensates publishers when their content is used inside AI Overviews, AI Mode, and Gemini answers. First spotted in April 2026 and now confirmed with more detail, the pilot was reported across Search Engine Journal, Search Engine Roundtable, and Marie Haynes' SNYCU newsletter. Google's approach is being compared against competing models from Cloudflare and Microsoft, which are also developing frameworks for paying publishers when AI systems access their content — creating a nascent but significant ecosystem of AI content compensation models.

Digiday separately reported that Google has framed its version as a "pay-value AI licensing program." The outcome of these competing models could reshape the economics of content creation on the open web.

Google is testing a "Dive Deeper" button in Google Discover that, when clicked, does not take users to a publisher's article — instead it displays an AI-generated short topic overview with links to related content. The feature is launching first with video content. Reported by both Search Engine Journal and Search Engine Roundtable, the experiment represents a further compression of direct publisher traffic from Discover, mirroring the pattern already seen in standard search results with AI Overviews.

Separately, Google's desktop Discover feature — announced and partially rolled out in 2025 — has quietly disappeared from Google.com and is no longer accessible, without any official explanation from Google.

In a significant concession, Google has publicly admitted that Search Console's current reporting does not adequately reflect AI search positioning data for SEO purposes. Google's John Mueller followed up by stating that position and impression tracking for AI Mode and AI Overviews in GSC will "evolve over time" as the products themselves evolve — signaling that the current data should not be used to draw firm conclusions about AI search visibility.

Reported by Search Engine Journal and Search Engine Roundtable, this confirmation validates what the SEO community has been raising: there is a meaningful gap between actual AI search visibility and what GSC currently surfaces. The SEJ SEO Pulse roundup also flagged Mueller's statement that AI position data is simply "hard to report" at this stage.

It has been an unusually volatile period in Google rankings. Search Engine Roundtable's Barry Schwartz — who noted he had been quiet on update reports since late August — flagged an unconfirmed ranking fluctuation around September 4th that appeared to revert around September 13th, describing the pattern as "super weird." Then on September 15th, Schwartz reported a major new spike in ranking volatility, describing "a huge spike in chatter" in the SEO community and movement in tracking tools, speculating the start of a new, unconfirmed algorithm update.

By the September 18th weekly video recap, Schwartz confirmed that volatility had continued and intensified throughout the week: "something is up." Notably, the September 13th reversal coincides with reports that Google has become significantly more effective at blocking AI scrapers, search scrapers, and third-party rank-tracking tools since around that date — potentially distorting ranking tool readings and complicating the picture for SEOs trying to understand what's actually happening in the SERPs.

Google is testing a new text-link ad format within AI Mode where sponsored results visually blend with organic AI-generated responses. The ads look like standard AI Mode text links with anchor text but are marked with a "Sponsored" label above the response block. This format blurs the visual distinction between paid and organic content in an AI-native search context — a development advertisers and SEOs should monitor closely for click-behavior implications.

OpenAI is testing a new ad format called Sponsored Agents in ChatGPT, which allows users to start conversations with business-sponsored AI agents after clicking an ad. Reported by Search Engine Roundtable, this represents a fundamentally different model from traditional paid search — rather than directing users to a landing page, the ad initiates an AI-native conversation with a brand-controlled agent. The format signals a new frontier for search and discovery advertising that SEO and SEM professionals should track as a potential emerging channel.

Cloudflare launched two significant features this week for managing AI bot access. Bot Preference Sync automatically generates and manages robots.txt policies by AI crawler *category* rather than per-crawler rules — allowing site owners to, for example, allow GPTBot while blocking Bytespider with a single policy setting, without needing to enumerate every bot individually. A separate feature, Disallow AI Training, lets site owners block crawlers from using content for model training without inadvertently blocking Googlebot or other legitimate search crawlers. Google and Apple already honor this training opt-out; Bing's robots.txt support is still pending.

Both features, reported by Search Engine Journal, come as Google, Cloudflare, and Microsoft develop competing publisher compensation models — adding context to why these crawler management tools are becoming increasingly important infrastructure for publishers.

Google's John Mueller responded this week to a report about cross-domain canonical tags causing unexpected de-indexing, explaining how publishers can monitor for and catch site problems before they escalate into an indexing crisis. According to Search Engine Journal, Mueller's guidance centers on proactive monitoring rather than reactive fixes — underscoring the risk of misconfigured cross-domain canonicals silently removing pages from Google's index. SEOs working with multi-domain setups should review their canonical configurations in light of this.

TechSEO Vitals Issue #45 highlights a persistent crawlability problem with headless CMS and JavaScript-heavy frameworks: content that appears normally in a browser is often absent from the raw HTML source — the version crawlers typically see first. The newsletter recommends a quick audit: open three recent projects, view the raw HTML source (not the elements panel), and search for the first sentence a visitor reads. If it's missing, there's a crawlability problem that may be suppressing indexation.

Google rolled out several Search Profile updates simultaneously this week. The follower threshold for eligibility has been lowered to 10,000 (from a higher bar), opening the feature to a wider range of U.S. creators and websites. New account management tools now allow a single login to manage multiple profiles, and articles display in a redesigned format. Separately, Google quietly launched a Search profile badge that website owners can embed on their sites, allowing visitors to follow them directly from the page — a detail omitted from the main announcement, as noted by Search Engine Roundtable.

Search Engine Journal framed the expansion in the context of a broader publisher traffic crisis, noting that Google's push to make Search a social-follow destination is actively redirecting user engagement away from publisher websites.

Google is converting local knowledge panels — including Google Business Profiles — into AI Overview-style listings. The new format displays an "AI Overview" label at the top of the local listing and includes a "Show more" button that launches an AI Mode-style chat interface for follow-up questions. Reported by Search Engine Roundtable, this is a significant UX shift for local SEO, potentially changing how businesses appear and how searchers interact with local results.

Google has upgraded Search Live to run on Gemini 3.8 Live, its latest multimodal AI model, bringing enhanced audio and video search capabilities to a feature that has been globally available since March. Separately, Google is testing the addition of an AI Mode button directly inside the search results page's search bar — adding to its already broad placement of the feature across Google's surfaces, a clear signal of Google's continued push to route users toward AI Mode.

Google has added four new ad-related metrics to the Chrome User Experience Report (CrUX): Ad Count, Ad Density, Ad Weight (Network), and Ad Weight (CPU). Reported by both Search Engine Journal and Search Engine Roundtable, this expands CrUX's scope well beyond Core Web Vitals and introduces publisher-facing UX signals that may carry future ranking implications. Publishers should begin tracking these metrics now as they move from testing to production status.

Google has updated its official documentation for the Mediapartners-Google crawler, broadening its scope beyond AdSense to include additional ad-related services. Site owners using robots.txt directives to manage this crawler's access should review the updated guidance, as the previous documentation may no longer accurately reflect the crawler's behavior and purpose.

As part of its ongoing Digital Markets Act compliance, Google has removed free product listings and carousel formats from the European Economic Area. Simultaneously, Google has updated documentation to include local business queries within its two-result-unit format for the EEA — a format previously limited to directories and direct providers. The European Ecosystem Carousel, which promotes third-party vertical search providers, is now also live for weather, translations, and other verticals in EU SERPs. Local SEOs and e-commerce operators in Europe face a materially different SERP landscape than their global counterparts.

Several studies and frameworks this week shed light on how generative AI systems choose what to cite. A controlled test reported by Search Engine Journal found that source order in AI citations matters less than raw data implies, while structured content rewrites measurably shifted how citation credit was distributed — suggesting that content formatting has more influence over AI citation than position alone.

A separate audit of Reddit's AI search feature found it disproportionately surfaces formal, highly upvoted comments while deprioritizing experiential language — an important signal for SEOs treating Reddit as a citation and traffic source.

Lumar's GEO diagnostic checklist distinguishes between four potential failure points for content missing from AI results: candidate eligibility, semantic relevance, chunkability, and authority — noting that page-level relevance does not guarantee passage-level retrieval. Meanwhile, Chris Green writing for SEJ offers a practical retrieval check: paste a distinctive content snippet into a chatbot and ask for exact matches — if your URL comes back, retrieval isn't your problem.

Google rolled out a cluster of Merchant Center updates this week. AI performance insights are now generally available in Merchant Center, alongside a simplified Universal Commerce Protocol (UCP) integration process and guidance on how conversational and video attributes improve product visibility. Google is also testing cart transfer and checkout testing within the UCP hub, along with a YouTube ads beta.

On the API side, the Merchant Center API now supports uploading product reviews data sources — a previously stated limitation that has been removed. Google is also testing a UI change that links product and shopping results directly to merchant websites instead of opening an internal side panel, which could meaningfully affect e-commerce traffic from shopping queries.

Additionally, Google now requires ecommerce sites to label adult products in their feeds or markup to remain compliant with updated guidelines.

Search Engine Journal released its State of Search 2027 report, providing data-driven guidance for navigating the current AI-driven disruption. The report frames traffic declines in context — arguing that a decline in traffic does not automatically mean a decline in business impact — and directs SEOs toward investments with clearest ROI. The release coincides with a broader industry conversation about what metrics matter in an era where GSC data for AI search is acknowledged to be inadequate.

Nick LeRoy of #SEOForLunch argues this week that the SEO industry's persistently low barrier to entry is fueling a trust crisis, and makes the case for formal accreditation or professional standards similar to those in other professions. The argument is particularly pointed given that AI tools are lowering the barrier further, making it easier than ever for unqualified practitioners to present as experts. The piece touches on a recurring debate in the SEO community that is gaining renewed urgency.