top of page
Wissenswertes
Suche


Long: Study on ChatGPT Query Fan-Out
Key Takeaways: Long (Nectiv) analyzed ChatGPT Query Fan-Out via 8.500+ prompts across different industries in October : Total Number Of Searches: 2,648 Percentage Of Search Instances: 31% Average Number Of Searches: 2.17 Average Words Per Query: 5.48 A search triggers 1 to 4 fan out queries that are mostly 3+ words long with an average of 5,84 words Compared to Google (3,4 words per query) the ChatGPT fan-out queries are significantly longer and more long-tail The main n-gram
23. Dez. 20251 Min. Lesezeit


Seer: Google's CTRs are dropping - AIO is only one reason
Key Takeaways: A study by Seer Interactive found massive CTR drops from Google comparing September 2025 vs. June 2024 Paid CTR without AIO dropped by -32% Paid CTR with AIO dropped by -68% Organic/SEO CTR without AIO dropped by -41% Organic/SEO CTR with AIO dropped by -61% Google proclaims that AIO is a successful feature for users and Google is sending more traffic to the web - data throughout different industries cannot support those claims Green raises important points:
22. Nov. 20251 Min. Lesezeit


Mantica: Google AI Max is worse than other match types
Key Takeaways: Mantica shows data from running Google Ads AI Max for 4 month: AI Max lost against all match types , doubled costs for conversions Google responds to criticism on X (AdsLiaison): [AI Max is] designed to help you capture incremental conversions or conversion value , depending on your objective. It does this by expanding your reach to additional relevant traffic using broad match and keywordless matching in the backend. It pairs this reach with optimizing your a
20. Nov. 20253 Min. Lesezeit


Terrasi: OpenAI leaks ChatGPT Data to Publishers - CTR <1%
Key Takeaways: Terrasi shared data from a document from OpenAI that showed very low CTR in ChatGPT of <1% for a website with over 500.000 impressions and an average of 1.68% Salomon: " It's a monthly report for each Publishers for the TOP 1000 urls" "What is obvious is that the data model is not prompt/query oriented but based on the URL ." Critical review: Industry is quite important - for publishers there are likely multiple sources for the same information and the inform
18. Nov. 20252 Min. Lesezeit


Korenugin: Research on factors for citations in LLMs (ChatGPT & AI Mode)
Key Takeaways: SE Ranking analyzed over 50.000 keywords in the US, mostly focussed on eCommerce / electronics Most impactful factors: Total number of referring domains Total domain traffic Page trust ... Their concluded recommendations: "1️⃣ + 2️⃣ Grow 𝐃𝐨𝐦𝐚𝐢𝐧 𝐏𝐨𝐰𝐞𝐫 & 𝐀𝐮𝐭𝐡𝐨𝐫𝐢𝐭𝐲 + 𝐁𝐫𝐚𝐧𝐝 𝐏𝐨𝐰𝐞𝐫: Make your brand visible across a variety of channels — media, social, review platforms, and traditional SEO backlinks. 3️⃣ Invest in 𝐓𝐞𝐜𝐡𝐧𝐢𝐜𝐚𝐥 𝐒𝐄
16. Okt. 20251 Min. Lesezeit


Williams-Cook: Google PAA increasingly AI generated
Key Takeaways: According to AlsoAsked Data on English queries the share of AI generated People Also Ask in Google SERPs increased from 17,8% to 37,8% . According to my observations this is likely similar in German/Germany Thus, data points in Google Search Console and tracking tools likely affected as spots for direct mentions and links are taken over bei AI Overviews-like answers Update 12/25 : while the maximum was hit in October (41.4%) Google seemed to have doubled down
14. Okt. 20251 Min. Lesezeit


Kunz: Google AI Mode in Deutschland und EU
Key Takeaways: Nach der Ankündigung am 07.10. ist nun auch seit dem 13.10. der AI Mode bzw. KI Modus in Deutschland ausgerollt Mit der Integration in der Suchleiste wird der AI Mode prominent angeteasert Christian Kunz fasst es gut zusammen: " Das Ergebnis dürften deutlich weniger Klicks sein. Das zeigen aktuelle Studien. Dabei dürfte klar sein: Der KI-Modus wird bleiben, und Websitebetreiber und SEOs müssen damit zurechtkommen . Dafür gibt es verschiedene strategische Ansät
13. Okt. 20251 Min. Lesezeit


DEJAN: ChatGPT 5 likely not using Schema Mark-up
Key Takeaways : Dan Petrovic tested the use of Schema Structured Data mark-up on a webpage of his asking ChatGPT 5 to summarize it There was no evidence of Schema even reaching the model The answers with and without Schema were quite similar ChatGPT itself replied multiple times that the browsing tool does not provide any Schema structured data just plain text from the raw HTML Inspired by a previous experiment that found impact of Schema on former ChatGPT models -...
22. Aug. 20251 Min. Lesezeit


DEJAN: Chrome’s Chunking & Embeddings Engine
Dan Petrovic analyzed Google Chrome and found the exact chunking mechanism and embedding logic used, now being able to to browse,...
21. Aug. 20252 Min. Lesezeit


Metahan: Inside ChatGPT 5 Search
Key Takeaways: Metahan analyzed ChatGPT chats and found the core features and configurations of it The Reranking Model =...
19. Aug. 20251 Min. Lesezeit


Mutznik: Adobe ChatGPT Study (US)
Key Takeaways : Adobe conducted a survey of 1.000 people in the US, 800 consumers and 200 marketers regarding their usage of ChatGPT...
18. Aug. 20251 Min. Lesezeit


Metahan: Perplexity Ranking Patterns in Response JSON
Key Takeaways: Metahan analyzed the JSON Responses of Perplexity and identified Ranking Patterns " Success requires not just keyword...
4. Aug. 20252 Min. Lesezeit


Josh Blyskal (Profund): ChatGPT wechselt von Bing zu Google
Key Takeaways: Profound Studie: Starker Shift in den Quellen-Verweisen von ChatGPT Übereinstimmungen mit Google Suchergebnissen deutlich...
29. Juli 20251 Min. Lesezeit


Beus (SISTRIX): AI-Chatbots erfinden Quellen
Key Takeaways: AI-Chatbots erfinden bzw. halluzinieren Quellen Mehr als 1/3 der Quellen waren 4xx Seiten: Gemini 33,4% vs. ChatGPT 36,5%...
20. Juli 20251 Min. Lesezeit
bottom of page


