Georgian Court Hotel: what changed when AI could trust its facts
Georgian Court is a 180-room boutique hotel in downtown Vancouver. It joined ALFIE on July 1, 2026. These results come from two full audits of its website (July 8 and September 29, 2026), the website's own activity records, an AI answer test on October 9, 2026, and direct booking figures reported by the hotel.
Direct bookings
Figures reported by the hotel. Tracing which bookings begin with an AI answer is an ongoing effort that ALFIE continues with the hotel. One hotel's story, not a promise of results.
Every audit measure improved
- Overall AI readiness+24 points
How easily AI assistants can identify the hotel, check its facts and trust what they find
Jul 8 42% Sep 29 66%
- Experience+56 points
Evidence that real guests stayed and said so: reviews, awards and press coverage
Jul 8 29% Sep 29 85%
- Authority+32 points
How many respected outside sites vouch for the hotel, and how clearly the hotel's site points to them
Jul 8 50% Sep 29 82%
- Answer readiness (AEO, answer engine optimization)+30 points
How easily AI can pull a correct answer, such as room count, hours or price, straight from the site
Jul 8 60% Sep 29 90%
- Expertise+24 points
Whether content comes from people who know the subject, shown through named authors and specific detail
Jul 8 20% Sep 29 44%
- Trust+19 points
Clear contact details, a privacy policy, a secure connection and clear ownership
Jul 8 59% Sep 29 78%
- Fact ownership+18 points
How much of the hotel's information is stated by the hotel itself, in a form AI can read directly
Jul 8 45% Sep 29 63%
- Information gain+8 points
How much the content tells AI that it cannot find on any other hotel's site
Jul 8 56% Sep 29 64%
- Fidelity Tax+6 points
The extra effort AI spends working out what a hotel means when it meets vague words like "luxury" or "boutique". It improves as those words are replaced with precise, checkable facts
Jul 8 18% Sep 29 24%
| Measure | What it means | Jul 8 | Sep 29 | Change |
|---|---|---|---|---|
| Overall AI readiness | How easily AI assistants can identify the hotel, check its facts and trust what they find | 42% | 66% | +24 points |
| Experience | Evidence that real guests stayed and said so: reviews, awards and press coverage | 29% | 85% | +56 points |
| Authority | How many respected outside sites vouch for the hotel, and how clearly the hotel's site points to them | 50% | 82% | +32 points |
| Answer readiness (AEO, answer engine optimization) | How easily AI can pull a correct answer, such as room count, hours or price, straight from the site | 60% | 90% | +30 points |
| Expertise | Whether content comes from people who know the subject, shown through named authors and specific detail | 20% | 44% | +24 points |
| Trust | Clear contact details, a privacy policy, a secure connection and clear ownership | 59% | 78% | +19 points |
| Fact ownership | How much of the hotel's information is stated by the hotel itself, in a form AI can read directly | 45% | 63% | +18 points |
| Information gain | How much the content tells AI that it cannot find on any other hotel's site | 56% | 64% | +8 points |
| Fidelity Tax | The extra effort AI spends working out what a hotel means when it meets vague words like "luxury" or "boutique". It improves as those words are replaced with precise, checkable facts | 18% | 24% | +6 points |
Source: ALFIE audits of georgiancourthotelvancouver.com, July 8 and September 29, 2026. Scores are percentages of the points available in each category.
Experience nearly tripled, from 29% to 85%, because the hotel's existing reputation was written down in a form AI can read: its TripAdvisor Travelers' Choice 2026 award, 9.4 out of 10 Booking.com rating, WorldHotels Elite and BW Premier Collection memberships, and press in HOTELS Magazine, Hotel Management and Trip Reporter UK.
What AI assistants say about the hotel
On October 9, 2026, twelve questions about the hotel were put to ChatGPT, Gemini, Perplexity, Claude and Copilot, giving 66 answers in all.
- Correct every time: the address, room count and check-in and check-out times were correct in every answer to those questions.
- Named from a topic alone: asked about a women-only hotel floor and a live jazz club, all five assistants named Georgian Court, though the question did not mention the hotel.
- The hotel's own site as the source: shown as a source in 31 answers. ChatGPT did so in 13 of its 15 answers.
A one-day snapshot that will be repeated so changes can be measured.
Server records show AI reading the hotel's facts
- About 17,400
visits from six AI and search companies to the hotel's pages, July 14 to October 12, 2026
- About 5,100
of OpenAI's visits were ChatGPT reading the pages while answering people's questions
- 6 to 290
signed facts, each with a digital signature and an evidence link, July 8 to September 29
- 335
verified fact files on the hotel website, each opening correctly every time it was requested
- 11,342
times Meta's robot collected the fact files, every day for 13 weeks in a row
- Dozens
of pages read each week by Anthropic, OpenAI and Perplexity after a mid-August site update, up from one to three
| Company | Visits |
|---|---|
| OpenAI (ChatGPT) | 6,107 |
5,629 | |
| Microsoft (Bing) | 2,397 |
| Meta | 2,332 |
| Anthropic (Claude) | 660 |
| Perplexity | 318 |
Source: the hotel website's activity records (server logs), July 14 to October 12, 2026.
Industry recognition
Georgian Court's audit is used as a worked example in the AI Hospitality Alliance's AI-Ready Hospitality Data Framework, contributed by Donna Rougeau.