SEO Recap covering July 6, 2026

6 min read·11 stories

Daily summary of what matters in SEO, GEO, AEO, and AI search generated with Claude Code (beware of hallucinations)

Generative Engine Optimization (GEO) & Answer Engine Optimization (AEO)

The report finds AI referral traffic climbing but swinging sharply month to month rather than growing steadily.
Referral volume is shaped by factors like on-platform internal search and how well a landing page matches the query intent.
Claude is quietly gaining ground as a referral source alongside the more visible ChatGPT.

The author concedes you can't track every AI recommendation but can assemble a reliable picture of brand presence.
The approach relies on sampling prompts to estimate how often a brand surfaces in AI answers.
The goal is a repeatable visibility signal rather than a single precise attribution number.

The author built a personal "brain" using Google's Open Knowledge Format (OKF): markdown files with YAML frontmatter typed as concepts, entities, playbooks, references, and systems.
An index.md file lets an agent target relevant areas instead of running retrieval across the entire knowledge base, and markdown links form a connected knowledge graph.
An automated system checks Google's documentation daily and updates the relevant reference files when Google changes things like AI Answers or Search Console docs.
Playbooks let an agent (the author uses Google's Antigravity) draft client proposals and run post-update site-impact analysis, cutting a two-day analysis to a few hours.

Google's data team published the Open Knowledge Format (OKF) on June 13, 2026 as a directory of linked markdown files with YAML frontmatter, aimed at internal company knowledge (v0.1).
Unlike flat page-by-page markdown mirrors (as Cloudflare serves), OKF's markdown links create a graph that tells an agent how concepts relate, not just what each page says.
The author built an eight-file OKF bundle for the No Hacks site (brand, host, Machine-First Architecture, agentic web, AEO, llms.txt, WebMCP), finding the main work was deciding which concepts connect.
A parallel machine-readable layer carries a maintenance tax: it goes stale whenever the site changes, and today no agents actually read website knowledge graphs.
The author notes Google is split internally: Search called llms.txt speculative, Chrome added an llms.txt check to Lighthouse, and the data team shipped OKF.

Meta AI is rolling AI-powered search into Facebook, Instagram, and WhatsApp.
The piece argues Meta's built-in distribution across those apps is its main advantage over standalone AI search tools.
It positions Meta AI as a potential large-scale answer surface marketers will need to watch.

SparkToro found 68% of U.S. Google searches ended without a click in early 2026, up from 60% in 2024; Ahrefs revised its estimate of AI Overview click loss to the top result from 34.5% to 58%.
Seer Interactive found brand-cited Overview CTR dropped 61% quarter over quarter while actual clicks stayed nearly flat, and cited pages earn roughly 120% more clicks per impression than uncited ones.
A randomized field experiment showed AI Overviews cut outbound organic clicks 38% with no change in user satisfaction, suggesting the lost clicks weren't only low-value bounces.
The author recommends tracking branded query volume, direct traffic, reading depth, repeat visits, and conversions, and building Rand Fishkin's correlation dashboard instead of a single traffic KPI.
Practical advice: add layers AI can't reproduce (interactive charts, video, downloads), and don't retire pages on traffic alone since they may still be cited or drive branded demand.

Brodie Clark spotted an early live version of Google's "Further Exploration" section appearing at the end of AI Overview answers.
Google's Hema Budaraju previously said the section links to unique articles or in-depth analyses on different facets of a topic to extend user curiosity.
The feature stems from Google's May announcement of five new generative-AI ways to explore the web in Search.
Sachin Patel also spotted a Clinical Guide within AI Mode, though it may not be new.

A Redditor trying to fix name-entity disambiguation used an llms-author.txt file plus Cloudflare's Content-Signal header (ai-train=no, search=yes, ai-input=yes) in robots.txt.
John Mueller said Google uses neither llms.txt nor llms-author.txt, and that no crawler or LLM he knows of honors the Content-Signal directive, which he called a CDN invention that just adds bloat.
llms-author.txt has no official proposal or standard behind it, and Cloudflare's Content-Signal is formally a robots directive, not a header standard.
The author notes the real fix is off-page reputation: interviews, podcasts, and a broader web footprint, not a technical SEO tweak.

John Mueller said Cloudflare's Content Signals robots.txt directive has "no effects whatsoever for any crawler or LLM" and just adds bloat and maintenance.
He said Google uses neither llms.txt nor llms-author.txt and knows of no crawler or LLM (outside SEO tools) that honors them.
Cloudflare, used by about 21.3% of all websites as of January 2026, has set a September 15, 2026 deadline to block Training and Agent categories by default on ad-carrying pages while leaving Search allowed.
Barry Schwartz advises Cloudflare users to check their settings before the default change takes effect.

The piece argues AI search is changing which SEO tactics deliver the most impact.
It names three priorities to emphasize more and three to pull back on as answer engines reshape results.
The author frames this as reallocating effort rather than abandoning SEO fundamentals.

Microsoft Bing is testing an overlay that shows product images, descriptions, retailers, prices, price insights and history, and related products when a listing is clicked.
The test appears in some browsers but not all, per Barry Schwartz's replication.
Clicking through moves users into Bing Shopping and out of the web search results.
Google has offered similar product detail overlay screens for a while.

Published on July 7, 2026