WATERMELON GHOST
Menu
Sample audit · Luxury · Amsterdam

From afterthought to first recommendation

Luxury 5-star hotel — Amsterdam, Netherlands

From afterthought to first recommendation cover

A pilot audit of a Leading Hotels of the World property in Amsterdam — Michelin-starred dining, decades of brand authority — measured against 60 real traveler queries across ChatGPT, Claude, Perplexity, and Gemini.

GEO Visibility Score: 36.7 / 100

Status: Afterthought. Branded authority is high, but generic recommendation visibility is structurally absent across the major LLMs. AI confidently endorses the hotel to guests who ask about it by name — and almost never surfaces it to guests who don’t.

The four decision moments

MomentScorePattern
Discovery25.0Zero indexing of the hotel’s unique positioning language
Recommendation25.0Absent from generic “where to stay” queries despite stronger credentials than appearing competitors
Comparison38.3Perfect branded performance (100), weak generic (22.9) — known but not recommended
Trust85.4Strong validation authority for guests who already know the brand

Platform mention rates

  • ChatGPT: 20.0% (12 of 60 responses)
  • Claude: 33.3% (20 of 60)
  • Perplexity: 41.7% (25 of 60)
  • Gemini: 63.3% (38 of 60)

The spread is diagnostic. Gemini surfaces the hotel often because Google’s own index picks up structured content; ChatGPT can’t extract the same content because it’s not in the right schema for ChatGPT’s primary data source. The visibility problem isn’t content — it’s structure.

What we found

  • A competing flagship captured 57 displacements across 19 territories where the hotel should have been competitive.
  • 16 zero-mention queries clustered around recommendation moments — the exact point where booking decisions get made.
  • The amenities page covered four gap territories but wasn’t being indexed by AI platforms for recommendation extraction.
  • Dining was the exception: fine-dining queries surfaced the hotel on Claude, Perplexity, and Gemini — proof that when the content foundation is right, the platforms cite it correctly.

The core pattern

“Strong validation authority but weak discovery optimization — the hotel validates well but doesn’t surface for initial consideration.”

A structural indexing gap, not a content gap. The 90-day execution plan reorganized existing assets into schema and positioning language AI platforms can extract — and prioritized ChatGPT-specific fixes where the gap was widest.