SISTRIX reveals Google self-preference in German AI Mode citations

SISTRIX examined millions of AI-generated responses following Google's AI Mode launch in Germany and documented systematic patterns favoring Google-owned properties. The German SEO analytics company published its findings on October 31, 2025, the same day Google made the conversational search feature available across Germany and numerous European countries.

Johannes Beus, founder and managing director of SISTRIX, announced the analysis through the company's monthly newsletter. "In the short time since the launch of AI Mode, we have already been able to analyze many millions of responses and identify initial clear patterns," according to the company's October review. The analysis focused specifically on German-language queries and citation behavior within the new market.

YouTube dominates AI Mode citations at 40%

SISTRIX's analysis of German AI Mode responses revealed YouTube as the most frequently cited source at 40.08%. Google.com itself captured 31.69% of citations, establishing Google properties as the dominant reference sources within the system. The data collection covered responses generated during the initial period following the German rollout.

The company compiled a comprehensive ranking of the top 50 domains appearing in AI Mode citations. Wikipedia secured third position at 17.83% citation frequency. Amazon.de received 13.92% of references, while Reddit.com appeared in 6.21% of citations. eBay.de registered at 4.29%, and Otto.de captured 4.14% of source attributions.

"One good news upfront: Even in AI Mode, there are still links to sources and websites," according to SISTRIX's analysis. "The less good news: Google most likes to link to its own offerings." The company documented two distinct citation mechanisms within AI Mode responses. Direct hyperlinks appear embedded within response text. Source citations display through small symbols positioned at paragraph endings, connecting to underlying reference materials.

The top 50 domains in AI Mode show strong similarities to organically visible domains, SISTRIX noted. However, the analysis identified substantial shifts in relative positioning and citation frequency compared to traditional search results.

Marketing implications of AI search shift

SISTRIX characterized the AI Mode launch as beneficial for users but challenging for brands and website operators. The company stated the feature will further reduce click volume from search results. This assessment reflects patterns documented in markets where AI Mode launched earlier.

The decline in click volume occurs because users obtain answers directly within search interfaces. Publishers and content creators face revenue implications when total visitor numbers decrease, even if individual visitor quality improves.

"How the weighting of sources changes in the coming months and whether other systems like ChatGPT can catch up remains exciting to observe," according to SISTRIX's October review. The company discussed current developments during the first GEO Conference organized by Claneo in Berlin. Johannes Beus spoke with Erik Siekmann on the Marketing Transformation Podcast about the transition from search engine optimization to generative engine optimization.

The conversation examined how AI systems select and prioritize sources. The discussion explored whether competitive AI platforms might challenge Google's positioning advantage for its own properties. The podcast episode became available online in October.

Prompt Monitoring launches for brand visibility tracking

SISTRIX introduced its Prompt Monitoring tool to address measurement challenges created by AI search features. The company offers the capability at no additional cost to existing SISTRIX users during the beta phase. "To still measure whether your brand appears in AI responses, there is SISTRIX Prompt Monitoring - free for SISTRIX users in the beta," according to the October 31 announcement.

The monitoring system tracks brand mentions across ChatGPT, Perplexity, and Google AI Overviews. Users configure customized prompt sets reflecting their specific monitoring priorities. SISTRIX then executes these queries on a regular schedule and analyzes resulting AI responses.

The tool reveals which competing brands appear in AI-generated answers for monitored topics. This competitive intelligence exposes positioning within AI recommendations. Organizations can identify new sources that AI platforms reference for specific subject areas.

Sentiment analysis extends beyond basic positive-negative classification. The system examines how AI platforms discuss monitored brands, providing qualitative assessment of brand portrayal. "With sentiment analysis, you can not only recognize whether the AI speaks positively or negatively about your brand, but also how," according to SISTRIX's tool description.

The Plus subscription tier includes 300 daily monitored prompts across three AI systems. Larger subscription packages provide expanded monitoring capacity. The company structured allocations to accommodate varying organizational requirements. The system tracks what ChatGPT, Perplexity, and Google AI Overviews display for configured prompt sets.

AI functionality expands across SISTRIX platform

SISTRIX released multiple AI-related capabilities during the third quarter. The company announced expansions in its AI Chatbot Beta, API, and launched a Model Context Protocol server. These developments help organizations measure visibility across various search systems, detect changes faster, and improve workflow efficiency.

Twelve new API functions deliver AI-related information. The endpoints provide data on prompts, entities, sources, and visibility across different systems. Organizations can retrieve AI analyses automatically and integrate them into existing workflows. The complete function catalog appears in SISTRIX's API documentation.

The company discontinued desktop SERP data updates on September 22 following Google's removal of the num=100 parameter. The desktop Visibility Index was eliminated as a consequence. SISTRIX modified its data collection methodology to accommodate the restriction, implementing a decision framework based on search volume and relevance for determining ranking depth.

By September 17, daily visibility data became available again for Germany, Austria, Switzerland, Italy, Spain, France, the United Kingdom, the United States, Sweden, and Brazil. The Netherlands and Poland received daily visibility data on September 19.

IndexWatch Q3 documents major visibility shifts

SISTRIX published its IndexWatch Q3 2025 analysis examining the June Core Update and August Spam Update impacts. Jolle Lahr-Eigen, SEO expert at SISTRIX, compiled the quarterly review covering the period's most significant winners and losers.

Two updates defined the quarter. The June Core Update and the August Spam Update shaped visibility patterns throughout the months. The August Spam Update ran longer than typical spam enforcement initiatives. Google simultaneously removed the parameter for 100 search results, creating temporary measurement challenges for SISTRIX and competing analytics platforms.

Domain migrations caused numerous visibility fluctuations. Hey.car, bmel.de, and bmfsfj.de lost visibility because operators deactivated old domains prematurely or configured redirects incorrectly. "The recommendation remains clear: Redirects should remain in place for at least one year," according to the IndexWatch findings.

Streaming and television reference databases achieved substantial gains. Werstreamt.es, fernsehserien.de, and imdb.com improved through detailed content pages about series and episodes. Traditional television program guides including Prisma and TV Movie lost visibility during the same period.

Wiki sites and databases showed recovery from Helpful Content Update impacts. However, rankings remained volatile for these properties.

T-Online suffered the quarter's most significant visibility loss. The domain dropped more than 57 points in SISTRIX's Visibility Index. The decline affected all directories without clear causation. Technical configuration errors impacted some organizations. The German Cancer Society blocked a central directory in robots.txt, preventing Google from processing redirects to the new domain.

Technical methodology behind the analysis

SISTRIX's analysis examined AI Mode responses generated through Google's query fan-out technique. This methodology divides user questions into component subtopics and searches each simultaneously across data sources. Custom Gemini models designed specifically for search provide the natural language understanding capabilities processing text, voice, and image inputs.

AI Mode queries measure approximately three times longer than conventional search queries, according to data Google shared during earlier rollouts. This length difference reflects the conversational interface accommodating complex, multi-part questions.

The top 50 domains in AI Mode demonstrate strong correlation with organically visible domains while showing substantial positioning shifts. The analysis documented how citation frequency differs from traditional ranking patterns, particularly regarding Google property representation.

SISTRIX maintains tracking systems monitoring millions of search result pages daily. The company adapted these systems following Google's September parameter removal, implementing selective depth measurement based on keyword value. The approach balances data quality against increased operational demands.

The Prompt Monitoring tool operates independently from traditional search measurement. The system queries AI platforms directly, capturing responses and analyzing content for brand mentions, source citations, and sentiment indicators. This parallel tracking infrastructure addresses the fundamental difference between traditional search result monitoring and AI response analysis.

Timeline

• October 7, 2025 – Google announced AI Mode expansion to over 40 countries and territories
• October 13, 2025 – Google launched grouped advertisement labeling with sponsored results headers
• October 28, 2025 – Multiple tracking tools detected search ranking volatility
• October 30, 2025 – Amplitude introduced AI Visibility tool with competitive tracking
• October 31, 2025 – Google launched AI Mode in Germany and European countries; SISTRIX published analysis
• September 22, 2025 – SISTRIX discontinued desktop SERP data following Google's parameter removal
• September 14, 2025 – Google removed num=100 parameter affecting measurement providers
• August 31, 2025 – SISTRIX documented AI Overviews appearing for 17% of German keywords
• August 26, 2025 – Google announced August Spam Update
• July 28, 2025 – Similarweb launched GenAI Intelligence Toolkit for brand visibility tracking
• July 2, 2025 – Google extended AI Mode to Workspace accounts in the United States
• June 30, 2025 – Google announced June 2025 Core Update

Summary

Who: SISTRIX, a German SEO analytics company founded by Johannes Beus, analyzed AI Mode responses and launched Prompt Monitoring capabilities. Google made AI Mode available to German and European users.

What: SISTRIX documented that YouTube captures 40.08% of AI Mode citations in Germany while Google.com receives 31.69%, demonstrating Google's systematic preference for its own properties. The company released free Prompt Monitoring for SISTRIX users to track brand visibility across ChatGPT, Perplexity, and Google AI Overviews.

When: The analysis and announcements occurred on October 31, 2025, coinciding with Google's AI Mode launch in Germany and multiple European countries following the October 7 global expansion announcement.

Where: The developments affect Germany specifically, with broader implications across European markets receiving AI Mode access. SISTRIX operates from Germany and focuses analysis on German-language search behavior.

Why: The AI Mode launch fundamentally changes how users discover websites, reducing click-through rates as users obtain answers within search interfaces. SISTRIX created Prompt Monitoring because traditional search measurement tools cannot track brand visibility within AI-generated responses, creating measurement gaps for marketing professionals.