SEO Recap covering October 12, 2026: Google poisoning rank trackers and ChatGPT citation shifts

Built by Stephanie Chung·5 min read·7 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)

Across 2.4 million answers in March and 3.8 million in September, PubMed climbed from 64th to fourth among ChatGPT's most cited U.S. domains while WebMD fell from 18th to 79th, and Britannica, Investopedia, and the New York Post dropped out of the top 200.
Reddit was cited in 19.1% of U.S. ChatGPT answers in March versus 2.3% in September, and Wikipedia fell from 14.1% to 2.9%, though both remain among the most cited domains.
Ahrefs says ChatGPT seems to prioritize original sources (government health and research sites, regulators, hands-on product testers like RTINGS) over sites that summarize them.
Of 66.6 million free, logged-out GPT-5.6 responses in August and September, only 14.4% ran a web search, and searching responses averaged ~1.6 fan-out queries. Of site: searches hitting the ~160 most searched domains, 75.9% went to brand sites.
Build pages only you can create (original research, hands-on testing, your own product information) and audit which sources ChatGPT cites for your core topics whenever OpenAI updates its model.

Microsoft introduced Microsoft-Decision-1, a fast decision-scoring model available in Microsoft Foundry and through OpenRouter, built for routing, classification, prioritization, verification, and workflow control.
Microsoft says the model outperforms both LLMs and other decision models on latency and quality for structured decision tasks.
It is already being experimented with in Bing Search and Copilot, though not across all queries or users, per Bing team members on X.

Google often shows AI Overviews when Search cannot match specific pages or when it treats a query as a typo, sometimes replying that the message "might have been a typo or random keyboard input."
The behavior reflects Google's goal of answering any query, even gibberish, rather than returning no result or admitting it does not know.
Google is also labeling some older search tools and widgets as AI Overviews, per Barry Schwartz.

2 medium-severity vulnerabilities affect AIOSEO, used on over 2 million WordPress installations, and both can be triggered by unauthenticated attackers under specific conditions.
CVE-2026-100152 (6.5) allows arbitrary shortcode execution on versions up to 5.0.2 when a missing validation check runs do_shortcode, but requires AIOSEO breadcrumbs to render on a search results page.
CVE-2026-85492 (6.1) is a DOM-based XSS on versions up to 5.0.1.1, exploitable only when a user with the aioseo_manage_seo capability opens the SEO Preview panel on a crafted URL.
Both are patched. Update to version 5.0.2.1 or newer to be protected against both issues.

Google updated its Merchant Center low image quality help doc to say products with blurry, pixelated, or low-resolution images will continue to show in ads and free listings.
The prior wording warned such products may be prevented from showing until image quality improves, so the change removes that suppression.
Google recently set the minimum product image size requirement to 500x500 pixels, and poor images may still hurt click-through rates even while listings display.

Nozzle's Derek Perkins found that since September 2026 Google has been serving rank trackers normal-looking SERPs with swapped-out results, with 8.0% of SERPs coming back poisoned and detectable only by comparing each keyword's own history.
Google is taking a 3-pronged approach to block scrapers:
- Wide IP blocking targeting major residential proxy networks
- A significant increase in captchas
- Injecting bad data into suspect SERPs so tools scrape wrong results
In poisoned results social sites and forums fill the gap and tracked sites fall out of the top 10, with the effect passing within days.
Tool reliability varies. Glenn Gabe reported Ahrefs picked up a spam update drop accurately, Semrush lagged, and Sistrix missed it entirely. Barry Schwartz suggests Search Console trends and incognito spot-checks as fallbacks.
Schwartz believes the effort targets AI bots and SerpApi rather than SEO tools, but the data-poisoning tactic is the hardest for tools to detect and may leave them unaware they are reporting junk.

A Principal UX Lead job posting describes transforming Discover from a recommendation engine into a "proactive, hyper-personalized concierge experience" that anticipates user needs and helps them act on goals, aligning with Sundar Pichai's agentic, task-based vision of search.
The role calls for new AI-enabled feed formats and engagement mechanisms while "elevating human voices," which could still benefit publishers and creators.
Google plans deeper integration beyond Search, partnering across YouTube, Shopping, Chrome, and Android for a unified experience.
The posting cites a track record with youth demographics, signaling Google wants Discover to compete with social media feeds for younger, stickier audiences.

Published on October 13, 2026