Patterns we see in ChatGPT query fanouts

If your content only answers the literal question typed, you miss what the model actually searches. Those hidden queries are fanouts — and they're one of the highest-leverage GEO signals.

Why fanouts matter

ChatGPT blends scores across subqueries (Reciprocal Rank Fusion). Content that matches multiple fanout angles beats content that matches one. Cover comparisons, reviews, brands, and year-stamped variants when the category warrants it.

Words ChatGPT injects

Top additions include best, top, comparison, reviews, tools, software, features. Advice-style prompts trigger 'best of' reframes roughly a quarter of the time. Reviews get searched even when the user never asked.

Model differences

• Perplexity (~1.4 fanouts): often simplifies the query — weak optimization signal
• ChatGPT (~2.1): adds brands, comparisons, reviews
• Grok (~6.8): research-brief style, often with site:reddit / Wirecutter / G2 style targeting

What to do

• Inspect fanouts for your tracked prompts in Anny
• Rewrite pages to cover injected angles
• Keep high-citation URLs freshly dated
• Watch review-site portrayals — they leak into answers