Get a behind-the-scenes look at how ChatGPT collects up-to-date information to generate answers. Our new fan-out queries feature in the LLM Scraper endpoint of AI Optimization API reveals the real search queries AI models use to explore your topic.
What’s new
The new fan_out_queries array in the LLM Scraper endpoint responses contains search queries derived from the main keyword that ChatGPT uses to retrieve additional information from search results.
Example:
"fan_out_queries": [ "Albert Einstein biography", "Albert Einstein early life", "Albert Einstein scientific contributions" ]
How it works
When you make a request to the ChatGPT LLM Scraper endpoint, the response will now include a fan_out_queries field, complementing the existing arrays of search_results (all web search outputs the model retrieved when looking up information) and sources (which the model actually cited in its final answer).
Why it matters
Data from the fan_out_queries field helps you to:
Identify new terms to target and optimize for in organic search.
Optimize your visibility in ChatGPT responses.
Discover related keywords the AI model actively explores.
Visit our updated documentation and start leveraging new keyword insights today. To learn more about how this feature works and explore a few practical use cases, check out this Help Center article.