70,000 AI responses checked – What the Blinq analysis reveals about visibility in ChatGPT & Co.

In a recent study, the analytics company Blinq evaluated over 70,000 responses from large language models (LLMs) such as ChatGPT, Google AI Overviews, and Perplexity to find out which sources these AI systems use most frequently for their responses—and what that means for online marketers.

LLMs are AI models that have been trained with enormous amounts of text and generate answers independently on this basis. They do not rely on an index like a search engine, but combine training knowledge with current web sources.

Corporate websites clearly dominate

A key finding of the analysis: over 51% of all cited content comes from corporate websites, i.e., company-owned blogs, websites, and landing pages—so-called owned content. This shows that owned content continues to play a dominant role when it comes to appearing in AI responses.

In addition, 13.3% of citations came from review media, which are often used to summarize product or service information. A further 10% came from the trade press and user-generated content (UGC), while pure news sites accounted for only 8.9% of sources.

This means that own content forms the foundation of AI visibility because it is consistent, trustworthy, and structured in terms of content—factors that LLMs use to select their sources.

What role does social media play in AI sources?

The Blinq analysis shows that platforms such as Reddit, YouTube, and LinkedIn account for a significant proportion of the sources used by LLMs:

• Reddit leads the way with 15.33% of the URLs cited.
• YouTube follows with 14.34%.
• LinkedIn accounts for 6.09%.

Developing content that is frequently shared and discussed in such communities increases the chance that AI models will use it as a reference.

Together, review sites and user-generated content account for almost 23.3% of the sources used—an indication that LLMs prefer authentic, “human” content.

Which news sources are cited?

Traditional media are also used by AI models, albeit to a lesser extent:

The news sources considered in the analysis are dominated by Chip (34.2%), Utopia (10.7%), WELT (9.8%), and TechRadar (6.3%). This shows that news sites are cited, but more so where concrete, comprehensive answers are required – such as product comparisons or specialist knowledge.

Overall, LLMs draw on news and specialist media in almost 20% of cases. These figures show that content strategies are not only about quantitative reach, but also about the type and depth of content that AI models classify as trustworthy.

How to become visible in LLMs

What can you do to ensure that your content is included in AI responses?

• Focus your content clearly on specific questions and problems faced by your target audience.
• Structure your articles clearly with understandable subheadings and precise statements.
• Systematically expand your website as a central source of knowledge.
• Actively position yourself in relevant communities such as Reddit or LinkedIn.
• Use YouTube and other video formats to showcase your expertise.
• Focus on sustainable thought leadership rather than short-term trends.

Want to know more about this topic? Find out everything you need to know about LLMs in our magazine article!