The platform spread that signals technical debt
Independent boutique hotel — Providence, Rhode Island
A pilot audit of an independent boutique hotel in downtown Providence — strong neighborhood positioning, mixed review sentiment — measured against 60 traveler queries across ChatGPT, Claude, Perplexity, and Gemini.
GEO Visibility Score: 44.1 / 100
Status: Considered option. But the headline number hides a more interesting story: the gap between platforms is one of the widest we’ve seen in a hospitality audit.
Platform mention rates — the diagnostic chart
- ChatGPT: 21.7% (13 of 60)
- Claude: 33.3% (20 of 60)
- Perplexity: 80.0% (48 of 60)
- Gemini: 71.7% (43 of 60)
When a property surfaces on Perplexity 4× more often than ChatGPT, it’s not a content problem. The content exists — Perplexity’s live web crawl finds it. ChatGPT can’t extract it because the schema and structured data aren’t in the shape its primary data source needs.
The four decision moments
| Moment | Score | Pattern |
|---|---|---|
| Discovery | 50.0 | Average overall, but ChatGPT scores 0 here — total absence at category-level queries |
| Recommendation | 43.5 | Mentioned in 47 of 108 generic recommendation responses |
| Comparison | 37.5 | Weak — branded comparison only 62.5, with negative sentiment present |
| Trust | 46.9 | Mixed — strong branded mention rate, but sentiment shows 6 negative and 19 neutral responses |
The sentiment problem
Unlike most audits, this property has negative sentiment registering across multiple platforms:
- Trust queries: 6 negative, 19 neutral, 11 positive (branded responses)
- Comparison queries: 3 negative branded responses
- Claude trust queries: 1 negative
- Perplexity trust queries: 1 negative
AI platforms are surfacing reputation signals — review patterns, service complaints, comparison disadvantages — and reflecting them in answers. This is a reputation layer that has to be addressed in parallel with the indexing fix.
Where the property disappears
8 zero-mention queries clustered on neighborhood and comparison moments:
- “Where should I stay in College Hill for a hip experience?”
- “Best boutique hotel in Providence with rooftop terrace?”
- “Looking for upscale accommodations in Federal Hill area”
- “Best value upscale hotels in Providence compared to Hotel Providence”
The property markets itself as a Federal Hill / downtown story but isn’t surfacing for either neighborhood-specific query.
Who’s getting recommended instead
| Competitor | Mentions |
|---|---|
| The Beatrice | 127 |
| Hotel Providence | 125 |
| Renaissance Providence | 86 |
| Omni Providence | 77 |
| Aloft Providence | 31 |
The Beatrice and Hotel Providence each surface 10× more often than the subject.
The core pattern
Two problems, not one: structural indexing on ChatGPT and reputation sentiment across the board.
The execution plan sequenced ChatGPT-specific schema fixes (the highest-leverage technical move) ahead of content work — and routed the reputation findings to the property’s PR team in parallel.