Perplexity cites 3× more vendors per prompt than Claude

Across our 20 standardised cybersecurity prompts, Perplexity returned an average of 8.4 distinct company citations per prompt. Claude returned 2.7. ChatGPT sat at 4.1. Gemini at 5.8.

This difference is architectural. Perplexity uses real-time web retrieval (RAG) to ground its responses — pulling fresh content at query time. This makes it more sensitive to current indexed content and broader in its citation set. Claude and ChatGPT in standard mode draw from training data, which tends to concentrate around the most established sources.

For GEO strategy, this has a direct implication: Perplexity is the easiest LLM to break into for a new entrant, because it has the widest citation distribution. Claude and ChatGPT require deeper topical authority and longer corroboration trails to crack.

A multi-LLM strategy — not single-LLM optimisation — is the only durable approach to AI visibility.