85% of B2B Topics Have No AI Answer Owner Yet

The Big Picture

Google spent the week handing site owners more visibility into AI search while its own engineers poured cold water on the loudest "how to win at AI search" advice. Search Console's AI performance reports went global on August 31, and John Mueller followed with two reality checks: markdown files aren't actually what AI crawlers are asking for, and unstable sitemap URLs undermine AI citations the same way they undermine traditional indexing. Meanwhile, fresh traffic data confirms the shift everyone's felt: ChatGPT.com is now the ninth most-visited site in the U.S. (up 48% year over year), LinkedIn just cracked the traffic top 20, and Bing fell 50%. Google Ads, for its part, isn't waiting around: automatic migration of legacy Search campaigns to AI Max started this month.

Google AI Mode & AI Overviews

Search Console now shows you where AI features cite you, minus the clicks

Google's generative AI performance reports in Search Console, first tested with a subset of sites in June, rolled out worldwide on August 31. Every site now gets a dedicated view of when its URLs appear in AI Overviews, AI Mode, and Discover's AI features, broken out by impressions, pages, countries, devices, and dates. There's no click data, so it won't tell you if an AI citation drove a visit, only that you showed up. Google also shipped a matching opt-out control for publishers who'd rather exclude their content from these surfaces entirely (doing so forfeits any traffic or impressions from them too).

Link carousels keep expanding

Google extended its developing-topic link carousel, first added to AI Overviews in May, into AI Mode. For fast-moving topics, users now see a horizontal row of timely article links (with Preferred Sources highlighted) instead of a single AI-written summary. Preferred Sources itself has grown to more than 600,000 unique sites, up from roughly 345,000 in May, and enrollment runs through Search Console.

Key stat: AI Overviews now reach over 2 billion monthly users across 200+ countries.

• The new AI reports are a measurement upgrade, not a strategy on their own: use them to confirm exposure, then pair with a citation-tracking tool if you need to connect AI visibility to pipeline.
• Preferred Sources enrollment is free and worth checking regardless of whether carousels currently touch your query set.

Sources: Google Search Console AI performance reports rolling out globally, Google Brings Developing-Topic Link Carousels To AI Mode

GEO / AEO / AIO

Mueller: skip the markdown rewrite, check your logs first

Responding to the wave of "serve markdown to AI crawlers" advice that's spread alongside llms.txt, Google's John Mueller said that on his own test sites, the only crawlers claiming to accept markdown are SEO tools, not the major AI bots. His advice: log the Accept header on your server and see what's actually being requested before building a separate markdown layer for your site. It's a useful gut check for B2B teams weighing dev time against unproven GEO tactics.

B2B still isn't cracking AI Overviews, and citation doesn't equal revenue

The gap between ranking and being cited remains wide: B2B brands that rank well in traditional Google results appear in only about 3% of relevant AI Overviews, even though AI Overviews surface on roughly half of B2B-relevant searches. Ahrefs' correlation study across 75,000 brands found that off-site brand signals, specifically YouTube mentions, branded web mentions, and branded anchor text, predict AI visibility 2 to 3 times better than backlinks do, and earned media now accounts for an estimated 82 to 89% of all AI citations. Only 12% of URLs cited by AI overlap with pages ranking in Google's top 10, and 44% of B2B brands with strong Google rankings have zero ChatGPT visibility.

Semrush's research (cited alongside its new Spotlight conference announcement) adds a sharper number: customers arriving from AI search channels convert 4.4 times better than organic search visitors. The same research found only 15.2% of the 1,094 topic categories Semrush analyzed have a clear AI-answer owner, meaning roughly 85% of B2B topics are still up for grabs.

Key stat: AI-search customers convert 4.4 times better than organic visitors, but only 3% of relevant AI Overviews currently include B2B brands that otherwise rank well.

• Content updated within the last 30 days receives roughly 3.2 times more AI citations than stale content.
• Comparison pages, ROI calculators, and implementation guides are the B2B content types most consistently pulled into AI answers.
• Ahrefs' September most-cited domains tracker still shows YouTube, Reddit, and Facebook dominating AI Overview citations. LinkedIn is the lone B2B-relevant domain in the top 50, holding a steady 0.6% share, a reminder that video and third-party discussion out-cite brand-owned content by a wide margin.

Sources: B2B brands rank in Google but appear in just 3% of AI Overviews, Semrush is bringing AI visibility under the Spotlight, The 50 Most-Cited Websites in Google AI Overviews, September 2026, Search Console AI Reports Go Global, Mueller On Recovery

Traditional SEO & Organic Search

A year of traffic data shows where the web's attention actually went

Fresh Semrush Traffic Analytics data (pulled July 2026 versus the year prior) shows the top of the web looks stable but isn't. Google is up 10.72% year over year and YouTube is up 36.6%. Between them, plus Reddit, the top three sites now generate 54.3% of all top-10 traffic. The clearest single mover is ChatGPT.com, up 48.38% year over year to 1.09 billion monthly U.S. visits, now ranking ninth overall and ahead of both DuckDuckGo and Bing. LinkedIn climbed into the top 20 for the first time in this data set, up 15.48% year over year to 473.7 million visits, a rare non-chatbot, B2B-relevant winner in a list otherwise dominated by consumer platforms.

The pressure this creates for publishers is now showing up in staffing decisions: USA Today's parent company reorganised its audience team this month, telling staff directly that search traffic is under pressure, and that platforms are increasingly keeping user experiences to themselves instead of sending them back. It's one of the largest publishers yet to name search traffic decline explicitly as a reorganisation driver, and it's trading page-view volume for engagement and subscription metrics instead.

Mueller also flagged a technical SEO habit worth killing

Separately, Mueller warned that cache busting a sitemap, appending a fresh version parameter to child sitemap URLs so search engines refetch them daily, is a bad idea because it signals unstable canonical URLs. The same logic extends to AI citations: a citation pointing at an address that changes daily is a citation that breaks. Worth an audit if your CMS or sitemap tooling does this automatically.

Key stat: ChatGPT.com is now the 9th most-visited site in the U.S., ahead of Bing and DuckDuckGo. Average organic conversion rate across B2B SaaS sites is still up 23% year over year despite lower click volume.

• Falling organic click volume alongside rising conversion rate is the expected pattern right now, not a red flag on its own: informational queries are increasingly resolved inside AI Overviews before a click happens.
• If your CMS auto-appends versioned parameters to sitemap URLs, check whether that's creating the same instability Mueller flagged.

Sources: The big winners and losers in U.S. website traffic over the past year, USA TODAY overhauls audience strategy, citing search pressure, Mueller: sitemap cache busting is a bad idea

New Protocols & Standards

llms.txt gets a real version 2, but adoption still outruns proof

llms.txt shipped a version 2 update in August, adding link relations so a page can point directly to its own llms.txt or Markdown copy via HTML link or HTTP Link headers (rel="alternate" type="text/markdown" and rel="describedby"), rather than requiring crawlers to guess at a root-level file. A June W3C proposal is pushing toward formal standardisation: fixing the file's location at the domain root, adding a version header, and defining a strict Markdown subset for the body.

The catch, underscored by Mueller's comments this week, is that adoption by IDE coding agents (Cursor, Windsurf, GitHub Copilot, Claude Code) is real, but major AI search crawlers still don't reliably request markdown or fetch llms.txt. Princeton's GEO-bench research found sites with a well-structured llms.txt earned 23% more AI citations across ChatGPT Search, Perplexity, and Google AI Mode than matched sites without one, but that's a correlation, not proof that the crawlers themselves are reading the file.

AI content watermarking becomes an adjacent compliance issue

Separately, Anthropic's August 11 rollout of machine-readable watermarking in Claude's outputs, a response to Article 50(2) of the EU AI Act requiring detectable marking of AI-generated content, triggered a wave of "AI content is now traceable" takes. The technical reality is narrower: it's a statistical signature embedded in word-choice patterns, not a visible mark, and Anthropic says it doesn't affect output quality or identify individual users. For B2B teams using AI-assisted drafting at scale, it's a preview of the kind of provenance and disclosure questions that are likely to keep surfacing as more jurisdictions adopt similar rules.

• Treat llms.txt as low-cost and plausible-upside, not as a proven citation lever. The ROI case will get much clearer if the W3C standardisation effort succeeds and crawlers start actually requesting it.
• If you're investing developer time in an AI-crawler-specific format, log your Accept headers first, per Mueller's advice above, rather than assuming.

Sources: llms.txt Got a Major Update. Google Says Skip It Anyway, Anthropic AI watermarking: What it means for content and SEO

Google Ads Updates

AI Max's automatic migration is live

Google's timeline for folding legacy Search campaign types into AI Max is now in motion. As of September 1, Search campaigns using Campaign-level Broad Match or standalone Automatically Created Assets are being automatically upgraded to AI Max on a rolling basis through the month, carrying over existing brand inclusions and exclusions. Creating new legacy Broad Match or ACA campaigns has already been blocked since August 3 across the Ads UI, Ads Editor, and API.

Dynamic Search Ads get more runway: Google will send in-account notices starting this month, reminder notifications in January 2027, and won't automatically migrate DSAs until February 2027. Developers using the Google Ads API should note that versions released after September 1 drop support for legacy Broad Match and ACA entities, with older API versions sunsetting around September 2027.

Key stat: Automatic AI Max migration for Broad Match and ACA campaigns began September 1. DSA migration is scheduled for February 1 to 28, 2027.

• If you're running Campaign-level Broad Match or ACA, this migration is happening automatically this month regardless of preference. Review account structure and asset groups now rather than after the fact.
• DSA advertisers have until February 2027, but Google's direction is clear: audit dependencies now instead of waiting for the forced migration.
• For B2B lead gen, AI Max is the closest replacement for what DSA did on long-tail, intent-driven queries. Confirm negative keyword lists and landing-page mapping carried over correctly post-migration.

Sources: Google sets AI Max migration timeline for Search campaigns

Suggested Actions This Week

• Check Search Console's new AI performance reports for your site to see where you're already appearing in AI Overviews, AI Mode, and Discover, and decide whether the generative AI opt-out control is right for you (it isn't, for most B2B sites chasing visibility).
• Before building a separate markdown layer for AI crawlers, log your server's Accept headers for a few days and see what's actually requesting it, per Mueller's advice.
• Audit your sitemap tooling for cache-busting or auto-versioned URLs. Stable canonical URLs matter for both traditional indexing and AI citation durability.
• If you run Google Ads Search campaigns on Campaign-level Broad Match or Automatically Created Assets, review your account now, the automatic AI Max migration is already underway this month.
• Pull a citation check across ChatGPT, Gemini, Perplexity, and Google AI Mode for your top three branded and category queries. With roughly 85% of B2B topic categories still lacking a clear AI-answer owner, there's real room to move.