StephEO

SEO Recap covering June 25, 2026: Google details AI controls as most bot traffic proves spoofed

Built by Stephanie Chung·7 min read·10 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)

AI Search Optimization Stalls on Internal Buy-In, Not Tactics
Source: Search Engine Journal

  • Wix's Crystal Carter distinguishes memory (what an assistant infers passively) from personalization (what users actively declare), arguing both can be shaped through owned-channel signals.
  • An iPullRank experiment running identical prompts across 3 accounts with different connected data produced visibly different AI Mode answers, including one naming a hypothetical child in a streaming recommendation.
  • Carter notes the average Google query runs 3-4 words while the average ChatGPT opening prompt runs ~103 words, making the case for FAQ-style, narrowly specific content over broad landing pages.
  • Tinuiti's Jen Cornwell reframes the problem as change management, citing Rogers' diffusion math: innovators (2.5%) plus early adopters (13.5%) reaching a 16% combined tipping point makes internal adoption self-sustaining.
  • The takeaways:
    • ship one FAQ-style proof-of-concept piece
    • identify the 16% of stakeholders already sold before pitching the full room
    • assign team roles (Sponsor, Skeptic, Catalyst) ahead of budget conversations

Weekly Martech Roundup Features a Wave of AI Visibility Tools
Source: MarTech

  • Zeta Global and Palantir announced a partnership to rearchitect Zeta's Data Cloud on Palantir Foundry, which Zeta's CEO said could generate more than $100 million in annual revenue.
  • Several new tools target generative engine optimization: Citelens tracks how often brands are cited or recommended across ChatGPT, Claude and Perplexity, while Contentful's Palmata and Efficiently Connected's SurfaceGX monitor brand reputation and source links inside AI answers.
  • Nudge raised $1.1 million in pre-seed funding and launched an Agentic Commerce Platform that measures product visibility at the SKU level inside engines like ChatGPT and Gemini.
  • DISQO's AI Search Lift uses a consented panel and exposed-versus-control testing to measure whether ad campaigns drive brand exploration inside LLMs and conversational search.
  • Aimzer's AI Visibility Framework and SurfaceGX's share-of-voice reporting both track citation presence and how often domains serve as source material for AI-generated answers.

AI in Search / AI Overviews

Google Publishes Detailed Docs on Search Generative AI Controls
Source: Search Engine Roundtable

  • Google released a help document explaining its new control over whether site content appears in AI Overviews, AI Mode and Generative AI features in Discover, in effect as of June 17, 2026.
  • 3 options exist: include (default), exclude, or inherit from a parent property. Child properties inherit parent settings but can override at the child level.
  • Excluding a site removes its links and grounding from AI features and forfeits any impressions or traffic, but Google says the control isn't used as a ranking or inclusion signal elsewhere in Search.
  • Content is excluded within one to 2 days of the control going live, though caching and propagation may take longer. Currently only UK sites support the control.
  • Glenn Gabe flagged the documentation, emphasizing the inheritance behavior as the section worth reading carefully.

Google Tests AI Overview Button Linking to Web-Only Results
Source: Search Engine Roundtable

  • Google is testing a button inside AI Overviews that takes users directly to the web-only tab, which strips out AI features and other verticals to show only standard web results.
  • Sachin Patel spotted the test, with a button labeled "See more commercial ramps" that routes to the web tab.
  • The web-only tab contains no AI search features, offering a path back to traditional links from within an AI Overview.

Mueller Clarifies When AI Features Count as a Search Console Impression
Source: Search Engine Roundtable

  • John Mueller said an impression in the new GSC AI performance reports requires a link to your site being shown in AI Overviews or AI Mode.
  • A linked page on your site counts as an impression, but Mueller is unsure whether a bare favicon would qualify as a link.
  • If a link needs to be "activated" to appear, it only counts as an impression when users perform that action.
  • The logic mirrors Google's existing Performance report documentation and applies as access to the AI performance reports expands.

Google Tests Autocomplete in AI Mode's Ask Anything Box
Source: Search Engine Roundtable

  • Google is testing autocomplete suggestions inside the Ask Anything follow-up box in AI Mode, surfacing suggestions as users type a follow-up question.
  • Brodie Clark spotted the feature, the first sighting of the new autocomplete display (announced at I/O) appearing within AI Mode.
  • Previews shown at I/O had autocomplete appearing in the global search bar with colored suggestions, hinting at wider rollout.

Technical SEO

Most AI Assistant and Googlebot Hits in Your Logs Are Spoofed
Source: Search Engine Journal

  • On the new CitationIQ.com site, only 6 of 33 requests carrying live AI-assistant names (ChatGPT-User, Claude-User, etc.) came from verified vendor IPs, an 81.8% spoof rate among checkable requests.
  • Of 799 requests claiming to be Googlebot, only 107 came from verified Google addresses, meaning ~87% were impersonators. Many spoofed assistant hits were credential scanners probing for .env.production, secrets.yaml and config.json.
  • Verify bots by matching the self-reported name against published IP range files (OpenAI, Anthropic, Perplexity, Google, Common Crawl) rather than trusting the user-agent string, using a 3 state outcome: verified, spoofed, or unverifiable.
  • Verified background crawlers were led by Anthropic's ClaudeBot at 166 crawls, ahead of Googlebot at 107, GPTBot at 46 and OpenAI's search crawler at 40, on a brand-new unpromoted domain.
  • Gemini can't be measured by name because Google bundles training, retrieval and crawl into one Googlebot plus the Google-Extended robots.txt token, which never fetches. The actionable takeaway is to run the same name-plus-IP check on your own access logs.

Google Search Console Page Indexing Report Stuck for Two Weeks
Source: Search Engine Roundtable

  • The GSC page indexing report has not updated in 14 days, frozen at June 11, 2026, per Barry Schwartz.
  • Performance reports were briefly backed up but appear resolved, while the indexing report delay is significant enough to flag at the 2 week mark.
  • Google indicates the delay is unrelated to the June 2026 spam update now rolling out. Debugging indexing issues will have to wait until the report is fixed.

Organic Search & Algorithm Updates

AWR: Desktop CTR Climbs While Mobile Slips at the Top
Source: Search Engine Journal

  • Advanced Web Ranking found desktop top-position CTRs generally rose over 2 quarters across 22 industries, while mobile declined at the #1 spot, reversing the broader downward CTR narrative for desktop.
  • On mobile, the #1 position dropped 2.20 percentage points with little change elsewhere in the top 10. Desktop gains showed up mostly below the third position.
  • Branded desktop searches gained in all top-10 positions (1.99-5.78 points), while unbranded queries saw a 3.07 point CTR drop at mobile #1.
  • The largest single swings were a 7.05 point desktop gain for first-ranked Family & Parenting sites and a 9.03 point mobile drop for first-ranked Law, Government & Politics sites.
  • AWR notes the data reflects its own tracking dataset, not direct algorithm changes, and assigns no causation to AI Overviews or ad layouts. The author warns against using blended or single-device CTR benchmarks.

Contentful Strategists on Where AI Should and Shouldn't Write
Source: Search Engine Journal

  • The argument is that AI generates by recombining existing content, so the more writing it handles, the more drafts sound like the pages already ranking for the same topic.
  • Speakers John Graham and Gabriel Dillon of Contentful place AI in research, SERP analysis, outlines and gap-finding, while keeping point of view, original data and expert angle human.
  • The session covers how to catch a draft that merely restates ranking pages before it hits the content calendar.
  • It also addresses reading engagement by reader segment to invest in formats and topics that perform rather than publishing on instinct.
Published on ·Last updated on
geoai searchtechnical seoorganic searchdigest