Rudzki: Peec AI Query Fan-Out study shows ChatGPT loves "best" and brands
- vor 5 Tagen
- 1 Min. Lesezeit
Aktualisiert: vor 3 Tagen
Key Takeaways:
Rudzki (Peec AI) analyzed 5 million query fanouts across ChatGPT, Perplexity, and Grok (April, 2026) and found:
Perplexity commonly just simplifies your query - not much insight for optimization (1.4 fanouts per prompt)
ChatGPT commonly adds brands, comparisons, and reviews - crucial for optimization for AI search (2.1 fanouts per prompt)
Grok runs a full research brief with year modifiers and site: operators - the richest signal of all if you want to optimize for AI search (6 fanouts per prompt)
ChatGPT
secretly adds "best" to recommendation queries for advice-style prompts ("Should I use X?", "Which CRM is right for me?") - which happens for 24.3% of fanouts. This is why "10 Best..." listicles dominate AI search results.
"Reviews" is the third most injected word ChatGPT hunts for reviews even when you never ask. For a Revolut sample, two sources shaped how AI described the brand: Glassdoor (4.87/5) and Sitejabber (1.3/5, with several fake-looking reviews). Smaller niche review sites slip under the radar.
LLMs love fresh content ChatGPT adds the current year to 5.44% of fanouts. Grok does it even more. Updating content already cited by AI can have positive impact on your visibility in AI search.
Grok
Takes query as research brief, starts broad and then narrows the query fanouts down
18.3% of fanouts add site: operator
10.5% of fanouts include site:reddit.com
Takeaways:
Anaylze and monitor your industry and search intents query fanouts
Make sure to be or become the authority for the respective content
Be present at authority websites and listicles
Sources:
