Wissenswertes

Key Takeaways: Segonzac analyzed a leak of Google Maps Geostore and found: 72 Geostore ranking signals 793 data source providers (along with mechanisms for provenance, priority, trust, and conflation = when several sources disagree) 446 local search intent types 50,998 Mapcore styles + 12,936 label styles = Mapcore decides what to show and what not 10,936 searchable Geostore declarations Google builds a canonical representation of the place that can incorporate data from mult

Key Takeaways: Metehan analyzed SSE (server-sent events) in the ChatGPT UI and found a detailed debug view of ChatGPTs retrieval system: For a single "best AI visibility tools" prompt, ChatGPT ran 5 search rounds, wrote 18 different queries (hidden queries), made 50 engine calls, pulled 228 results, fetched 223 URLs with selected chunks, and cited 16. The page body is not HTML. It is a markdown-like text render. That render is what gets cut into blocks of roughly 170 words an

Key Takeaways: Google introduces masked links with /goto-Redirect dropping another bomb on Rank Trackers and AI Chat competition It is a trap: While tool providers celebrate easy fixes and workarounds, I believe it is a trap: Now Google can identify more easily which users misuse its search engine = those sessions call all the redirect links from a SERP - before it could only see users or IPs doing lots of searches that could still look natural, now it actually sees who revea

Key Takeaways: Anthropic announced that all text from Claude models that launched on August, 2nd 2026, will have an embedded watermark to be identified under EU AI Act Claude will further provide documentation and tools in cooperation with partners to use for detecting their watermark Still, they also speak of several limitations How does it work? A Claude model will subtly shift word choices during generation using a secret cryptographic key, creating a statistical pattern t

Key Takeaways: On August 6th OpenAI changed the default model for ChatGPT to GPT 5.6 so that free tier users per default get access to GPT 5.6 Luna - eventhough just recently in July they've already changed it from GPT 5 mini to GPT 5.5 Salomon/OnCrawl analyzed logs and found ChatGPT User Bot jumped in hits when the new default GPT 5.6 was released and concluded Live Search is triggered more often for Free tier users Live Search relies on more Query Fanouts and generates more

Key Takeaways: Gabe reported an increase of ChatGPT ads based on a SE Ranking analysis by Deda and Tomko ChatGPT ads appear on 26% of commercial prompts, 61% of prompts regarding Pets 96.37% of brands that got ChatGPT ads displayed are not cited in the answer = only 3.63% of advertisers were also cited as a source in the answer above their ad YMYL: ChatGPT showed substantially more ads for YMYL topics than Google’s AI Mode. Ads appeared on 28.69% of healthcare prompts, compar

Key Takeaways: Kopp analyzed search patents related to search and AI search, adding a new patent on Search System Having Task-Based Machine-Learned Models The patent is a about a search system that breaks broad user queries into ranked subtasks using LLMs + specialized multi-task ML models — then surfaces, ranks and dynamically re-ranks content based on aggregated user interaction scores across the full task session. He outlined the process: Query Classification — is this a g

Key Takeaways: Segonzac analyzed the retrieval system of ChatGPT and gained highly interesting insights: Bright = web search engine but likely just Google Labrador = an in-house index, topped up with press feeds, open scientific repositories and partner platforms (it is not Bing: as titles are too long) The labrador snippet: 200 characters taken from the start of your page body (straight after your H1) and the page's title What to do with this. Your grounding budget in instan

Key Takeaways: Harpreet shows Ahrefs data on AI Overview citations of X.com that are increasing since March 2026 to July 2026 from approx. 250.000 to 900.000 citations As Ray raises concerns about how trustful and truthful X Posts are (and other social media platforms) Andell shows that the same is happening in Google Discover: Acceleration from the beginning of May (~38k), then a climb to ~65k by the end of the month, with the percentage visibility also doubling. After intro

Key Takeaways: Google introduced the first update to the recently introduced Open Knowledge Format (OKF) that adds a Trust Layer called OKF v0.2: new sources field: author, usage_count, last_modified generated: { by, at }: how the current content was produced, and when it last meaningfully changed. verified: [ { by, at } ]: a list of independent confirmations against the sources or the underlying resource; a human sign-off, a nightly finance process, or both. status moves a c

Key Takeaways: Konitzny (PeecAI) analyzed Query Fan-Outs and citation behavior of AI Search / LLM Chats and how they significantly shift over time In May Reddit citations as well as Query Fan-Outs with "reddit" increased and dropped in June again End of June Query Fan-Outs with "Official" (products / site / docs) jumped from 0.x% to 3.5% - in multiple countries --> government authorities, consumer protection sites and domains with "official" gain citation share In May the ope

Key Takeaways: Must watch: Petrovic outlined in a webinar from Edge of Search how AI Search Optimization works, what to focus on and why (AI) SEO experts have all the right tools to optimize a brands performance He compares monitoring model perceptions via prompts with human psychology and stresses how new tactics on top of traditional SEO can move the needle to incease citations, mentions and recommendations Why AI SEO? "When AI Search or Agents become personal assistants t

Key Takeaways: Segonzac clarified how websites can end up in the training data of a llm model, specifically that being crawled does not mean you are successfully finding your way into the training data. Layers of AI Chat Answers Training Data of the LLM model takes place in multiple steps: Crawling Extraction Deduplication Filtering Tokenization/Data mix/etc. Training Retrieval e.g. via Web Searches: Potential crawling hit Training Data Myths: “If an AI crawler hits my page,

Key Takeaways: SISTRIX reported changes in ChatGPT answer behavior when switching default model from ChatGPT 5 mini to ChatGPT 5.5 calling them "ChatGPT Core Update": 47% of cited domains changed within 48 hours in Germany German publishers and service brands won in visibility, while international aggregator lost Publishers Welt, FAZ, Bild, Chip, Computerbild are cited more similarly to their organic Google SEO visibility Garg reported a surge of ChatGPT Ads of +300x: see Wri

Key Takeaways: Pelham (Otterly) ran an Reddit Engagement Experiment: 9x more AI citations for the active community vs. the dormant one 6.9x more total citations than the median external subreddit 4x more Google keywords ranked 18x more monthly search volume covered Metehan (PeecAI) reported how Reddit is eating non-English AI visibility: We analyzed 64.77 million Reddit citations across 20 countries and 4 LLMs. In Sweden and Norway, over 70% of Reddit citations on Google AI O

Key Takeaways: Similarweb carried out a study on User Behavior in the US: Methodology: We tracked thousands of real user journeys across three industries: Finance, Travel, and Beauty - focusing on users who asked ChatGPT for an industry-relevant question and received a specific brand in the answer. We then followed those users for the seven days after the AI conversation, tracking whether they visited the website of the recommended brand or a direct competitor. To isolate gen

Key Takeaways: Google introduced new standards named "Open Knowledge Format" and "Agent Resource Discovery" OKF for Google should work as: a vendor-neutral, agent- and human-friendly standard for representing the metadata, context, and curated knowledge that modern AI systems need. Just markdown — readable in any editor, renderable on GitHub, indexable by any search tool Just files — shippable as a tarball, hostable in any git repo, mountable on any filesystem Just YAML front

Key Takeaways: Elvis emphasizes the importance of owning the orchestrator and the harness for complex AI workflows the orchestration, the harness, routing capabilities, dynamic artifacts/workflows, verifiers, ability to switch/route between agent backends, automations, the skills, and the MCP tools would be the absolute best defense for what happened with Fable this week. The potential downside: high maintenance, might be too costly, and migth be unsustainable The bigger you

Key Takeaways: Peham/Otterly analyzed 1.3 million LinkedIn AI Chat citations from ChatGPT, Perplexity, Google AI Overviews and AI Mode: Linkedin's AI citations rose +49.9% between January to May 2026 from 7.8% to 11.7%. People's posts get 91.7% of citations, rest goes to company posts. LinkedIn Pulse articles get cited far more than posts. More engagement /adding media doesn’t increase citation BDP / measures for increasing LinkedIn AI citations: Publish pulse articles, not j

Key Takeaways: Perplexity announced some great updates such as introducing Hybrid Agentic Inference and Search as Code Hybrid Agentic Inference enables you to split tasks between a local model running on your machine and frontier models in the cloud. This keeps private data on your device and maximizes token efficiency Only let Perplexity access a local folder and run a local model via a subagent, then switching to more powerful, expensive models in the cloud Search as Code c