11/19/2025
How to get Visible in LLM tools like ChatGPT, Claude, Perplexity and Gemini
• 💡 Query Fan Out is Key: Understanding the query fan out—where the LLM breaks a user's prompt into simpler search queries—is the direct link between SEO and LLM visibility.
• 🔄 Managing the Drift: The constant change in the queries LLMs use (drift) means that websites ranking for the most variations of a topic are typically cited, potentially requiring publishing the same articles under different keyphrases to target these various drift terms.
• 📈 Real-Time Ranking: LLMs fetch documents in real time; if you update and re-index a page in Google, you can see it reflected in LLM results shortly thereafter.
• 🔍 Finding Hidden LLM Searches: Low-DA sites often see very deep, long-tail queries (e.g., those using "evaluate") ranking in the top 10 in Google Search Console but with zero clicks; these are often LLM fan-out queries.
• 🖱️ Measuring LLM Traffic: Clicks from LLMs are registered as referral traffic (not search traffic) in tools like Google Analytics 4 (GA4) or Looker Studio.
• 🗣️ Guessing Prompts: A strategy to uncover potential user prompts is to take a screenshot of the landing pages receiving LLM referral traffic and ask an LLM (like Perplexity) to guess the user prompts that led to those pages.
• 🛑 Debunked Myths: LLMs are not search engines. The claims that they prefer schema or E-A-T (Expertise, Authoritativeness, Trustworthiness) are often misinformation. For most content (99%), schema is not necessary for ranking and does not replace the quality of the content itself.
• 🗺️ Focus on Topical Authority: Instead of focusing on schema, SEOs should prioritize establishing topical authority and analyzing Google Search Console data to uncover long-tail keywords and variations that may have high volume, even if initial rankings are low (e.g., position 76).
• 💻 Finding ChatGPT's Queries: You can find the exact language ChatGPT searches by using Chrome DevTools (Network tab), filtering the results based on the unique string, and looking at the response for "query"
E859: David Quaid returns to explain how large language models surface webpages and how you can measure and influence your visibility in tools like ChatGPT, ...
