Recently, ERGO Innovation Lab and ECODYNAMICS conducted a study focusing on how insurance-related content is presented in AI-driven search results.

The study analyzed over 33,000 AI search results and 600 websites, exploring the preferences of large language models (LLMs) like ChatGPT when processing this content. The findings revealed that LLMs tend to favor content that is easy to read, well-structured, and trustworthy—characteristics closely related to traditional SEO strategies.

Code Internet (1)

Image source note: The image was generated by AI, and the image licensing service provider is Midjourney.

In terms of content presentation, modular content, especially content presented in question-and-answer format, is more likely to appear in AI-generated answers. This aligns with the principles of classic SEO, where clearly linked and well-structured content is more favored by search engines. Additionally, the study compared the "hallucination" rates of different models, revealing that ChatGPT had the highest error rate, with nearly 10% of its answers containing inaccurate information, while you.com provided more reliable results. These findings are particularly relevant to insurance-related queries.

Through this study, we can see that although AI technology continues to advance, there remains a similar logic and principle between LLMs and traditional SEO in terms of search content performance. This not only provides valuable insights for content creators but also offers a reference for businesses to adjust their digital marketing strategies.

Key points:

🌐 LLMs prefer content that is easy to read and well-structured, which aligns with traditional SEO strategies.  

📝 Modular and question-and-answer format content has an advantage in AI-generated answers.  

🔍 ChatGPT has the highest error rate, close to 10%, while you.com is more reliable.