The Rented Ground

The Big Picture

Google's August 2026 spam update hit harder than any update this year, reshuffling rankings across 20 industries just as AI Mode crossed 1 billion monthly users and published its first full year of usage data. Meanwhile, an unannounced backend change at OpenAI wiped out most of Reddit's ChatGPT citation share overnight, a stark reminder that AI answer visibility can vanish as fast as it appears, and that B2B teams need direct, owned-content citation strategies rather than reliance on any single third-party platform.

Google AI Mode & AI Overviews

AI Mode turns one, and query behaviour has fundamentally shifted

Google shared its first full-year usage report for AI Mode, tied to I/O 2026 announcements. The average AI Mode search is now three times longer than a traditional query, and both short and long queries are growing. For B2B content teams, this confirms that conversational, multi-part questions are now a mainstream entry point, not an edge case.

Key stat: AI Mode has surpassed 1 billion monthly active users. Follow-up queries rose more than 40% month-over-month in the U.S., and queries have more than doubled every quarter since launch.

• AI Mode is now sometimes powered by Gemini 3.7 Flash.
• More than 1 in 6 AI Mode searches are multimodal (voice, image, or video).
• Google added a save function for AI Overviews (rolling out to U.S. English queries since August 15) that reproduces the same result set on repeat searches.
• A simplify the language button now lets users request a plainer-English rewrite of an AI Overview.
• Google rolled out link carousels for developing topics in AI Mode (following the same feature already live in AI Overviews): a horizontal row of article links surfaced between AI-generated response sections, prioritising users' Preferred Sources.
• Google is dynamically expanding AI Overviews in length for some queries.

Sources: Google Search's I/O 2026 updates, Google AI Mode Usage Data After One Year, Google adds link carousels for developing topics in AI Mode, Google is dynamically expanding AI Overviews for some queries

GEO / AEO / AIO

Reddit's ChatGPT citation collapse is this week's GEO case study

An unannounced backend change at OpenAI caused Reddit's share of ChatGPT Search citations to fall from 3.83% (July 18 to Aug 7) to just 0.52% (Aug 14 to 17), an 86.4% drop in weeks. Google's AI Overviews and AI Mode showed only a gradual decline in Reddit citations over the same period, suggesting the change was specific to ChatGPT's retrieval pipeline rather than a broader shift away from user-generated content.

Key stat: Reddit's ChatGPT Search citation share dropped 86.4% in under a month. Its Google AI Overviews citation share slipped only from around 2.5% to 2.1% over the same window.

For B2B marketers, the lesson is direct: platforms you don't control can change your AI visibility overnight. Diversify citation sources across owned content, structured data, and multiple AI engines rather than depending on any one third-party site.

• Semrush's expanded 2026 AI Visibility Index analyzed 126 million U.S. AI search prompts (January to April 2026). Companies that fully integrate SEO and AI-visibility workflows report far better outcomes than those managing them separately (81% saw increased AI-driven traffic or leads versus 36%).
• A 50-company B2B SaaS benchmark running 1,400 buyer-intent prompts across ChatGPT, Perplexity, Claude and Gemini found an 87-point gap between the top AI Presence Score (89) and the bottom (2), despite both companies having active marketing teams.
• Ahrefs data shows AI search visitors convert at roughly 24 times the rate of organic search visitors (12.1% of signups from just 0.5% of total visits).
• Content that performs best for AI citation continues to be original research, data-backed guides, comparison content, and clearly attributed expert quotes, not promotional copy.

Sources: Reddit nearly vanishes from ChatGPT citations, Reddit fades from ChatGPT citations, Semrush 2026 AI Visibility Index, The Complete Guide to AI Visibility for B2B SaaS

Traditional SEO & Organic Search

The August 2026 spam update caused unusually large ranking swings

Google's third spam update of 2026 rolled out August 18 to 21 (2 days, 16 hours) with no new policy types announced, but the ranking impact was well above normal. SE Ranking measured an 82% increase in top-10 URLs falling out of the top 100 compared to a non-update baseline, alongside a 12% jump in URLs newly reaching the top 3 that hadn't previously cracked the top 20.

Key stat: Search disruption from the August 2026 spam update spanned 20 industries, with an 82% increase in top-10-to-out-of-100 ranking drops versus a normal period.

• Organic search remains the largest B2B revenue channel: search drives 76% of all B2B site traffic, and organic search generates 44.6% of total B2B revenue, still more than half of inbound leads.
• Despite AI search growth, Google still sends roughly 345 times more traffic than all AI platforms combined, and 99% of AI users still use traditional search engines too.
• New traffic analysis shows a widening gap between YMYL and B2B tech content performance, pointing marketers toward top-of-funnel strategies as the more resilient play going forward.
• A 60,000-query listicle study found ranking position increasingly decoupled from click volume: ranking well no longer guarantees traffic.

Sources: Google's August 2026 spam update hit rankings harder than normal, Google Finishes Rolling Out The August 2026 Spam Update, 60 B2B SEO Statistics to Shape Your Strategy 2026

New Protocols & Standards

llms.txt hits version 2, and W3C standardization is moving forward

The llms.txt spec shipped v2 on August 10, 2026, incorporating two years of real-world usage feedback. The update adds HTML link and HTTP Link header discovery mechanisms (rel="alternate" type="text/markdown" to point to a page's Markdown copy, and rel="describedby" to point to the llms.txt covering it). Separately, a June 2026 W3C draft proposes formalising llms.txt further: a fixed domain-root location, a versioning header for crawler compatibility, and a strict markdown subset for consistent parsing.

Key stat: Princeton's GEO-bench research (May 2026) found sites with a well-structured llms.txt earned 23% more AI citations across ChatGPT Search, Perplexity and Google AI Mode versus matched control sites without one, though Google has publicly said it doesn't currently use the file for ranking.

• Adoption remains modest (an estimated 5 to 15% among tech and documentation sites), but it's treated as a baseline expectation among AI-native and developer-tool companies.
• Structured data's role is shifting from a SERP rich-result trigger to an AI trust and entity-verification signal. Google retired FAQ rich results on May 7, 2026, closing out a nine-year feature, and continues to de-emphasise FAQ, Review and How-To markup on pages where it isn't the primary content.
• Guidance for 2026: schema should match the primary topic of the page it's on. Mismatched or peripheral markup is now actively discounted rather than just ignored.

Sources: llms.txt Got a Major Update. Google Says Skip It Anyway, What the June 2026 W3C Proposal to Standardize llms.txt Means, Schema Markup After March 2026, Google Structured Data Changes 2026: What B2B SEOs Need to Know

Google Ads Updates

The AI Power Pack is now the default full-funnel playbook

Google is pushing advertisers, B2B included, toward running Demand Gen, AI Max for Search, and Performance Max together as a coordinated full-funnel system rather than choosing one campaign type. Demand Gen builds awareness and lets advertisers target by audience signal (useful for job-title and company-size targeting across YouTube, Gmail and Discover). AI Max for Search captures new intent. Performance Max handles final conversion using Google's own AI-driven bidding.

• 2025 to 2026 updates have narrowed the gap between Demand Gen and Performance Max, with both now sharing more inventory, reporting, and AI-driven capabilities.
• Demand Gen's auto-generated video feature now turns existing assets (images, copy) into ready-to-run video ads without a production team, which is relevant for B2B teams without in-house video capacity.
• Recommended B2B posture: use Demand Gen for top-of-funnel research-stage targeting by role and firmographic signal, and reserve Performance Max for bottom-funnel conversion where first-party conversion data is strong enough to guide the algorithm.

Sources: Your guide to Google's Power Pack: AI Max, PMax, Demand Gen, Every Google Ads Change That Matters in 2026, Demand Gen vs. Performance Max: B2B Funnel Strategy

Suggested Actions This Week

• Audit rankings against the August 18 to 21 spam update window. If you saw sharp drops, check for thin, AI-generated, or over-optimised content rather than assuming a manual penalty.
• Don't over-invest in any single third-party citation source (e.g. Reddit) for AI visibility. The ChatGPT citation collapse shows how fast that can change. Prioritise owned, structured, citable content instead.
• Audit schema markup for topic alignment. Remove or fix FAQ, Review and How-To markup on pages where it isn't the primary content.
• Evaluate publishing an llms.txt file (v2 spec) for key documentation and resource pages, especially if you compete in AI-native or developer-facing categories.
• Rewrite top commercial and comparison pages with lead-first, direct-answer paragraphs. AI Mode's one-year data confirms users are asking longer, more specific questions that reward front-loaded answers.
• If running Google Ads, test a coordinated Demand Gen, AI Max and Performance Max structure rather than a single campaign type, using firmographic and job-title targeting in Demand Gen for top-of-funnel B2B prospecting.