Last updated July 2026
AI-sourced travel traffic grew 3,500% year over year in July 2025, according to Adobe Digital Insights data covering more than eight million visits to U.S. travel sites. That growth is now a permanent shift in how travelers research trips. The problem for most hotels is that very few of them are capturing any of it.
The Hotelworld AI figure is the clearest statement of that problem: 84% of hotel inventory is invisible to the travelers who start their search in an AI engine. This guide explains how to find out which side of that line your property is on, and what to do about it.
What “AI invisible” actually means for a hotel
When a traveler types “best boutique hotels in Lisbon with a rooftop pool” into ChatGPT or Perplexity, the engine generates a response from a mix of training data, retrieval sources, and editorial content. Your property either appears in that answer or it does not. There is no position 6 to climb from. You are in or you are out.
AI share of voice for a hotel is the percentage of relevant prompts, across a defined set of AI engines, in which your property is mentioned by name. A property with 40% share of voice appears in four out of every 10 searches for its city and category combination. A property with 0% share of voice is invisible to that entire channel.
Unlike Google search, AI engines do not return the same answer to everyone. According to an Ahrefs study of 15,000 queries, only about 12% of URLs cited by AI assistants also appear in Google’s top 10 results for the same query. Ranking on Google does not protect you from AI invisibility. You need a separate measurement.
Step 1: Build your city-level prompt set
The audit starts with a prompt set that reflects how real travelers phrase their searches. Do not use your hotel’s name in the prompts. You are measuring unprompted category inclusion, not branded recall.
Structure prompts around three variables:
- City or neighborhood (“in Lisbon,” “near the Eixample in Barcelona”)
- Property type or category (“boutique hotel,” “design hotel,” “small luxury hotel”)
- Amenity or occasion (“with a rooftop pool,” “for a honeymoon,” “pet-friendly”)
Combine them into 20 to 30 distinct prompts covering the categories your property competes in. Examples:
- “Best boutique hotels in Lisbon with a rooftop”
- “Small luxury hotels near the Marais Paris”
- “Pet-friendly design hotels in Barcelona with a spa”
- “Romantic boutique hotels in Lisbon under 300 euros per night”
Run each prompt on ChatGPT, Google AI Overviews, and Perplexity. Run each prompt five times per engine before drawing any conclusion. AI engines are probabilistic: a single response is one sample from a distribution, not a reliable read on where you stand.
Step 2: Collect and score the responses
For each run, record:
- Which properties are named
- Where in the response your property appears (first, middle, or end of the list)
- Which sources the engine cites, if visible (particularly on Perplexity)
- Whether your property’s description is accurate
Build a simple scoring table. Here is the format that works for a city-level audit:
| Prompt | ChatGPT (5 runs) | Google AI (5 runs) | Perplexity (5 runs) | Your property present (Y/N) |
|---|---|---|---|---|
| Boutique hotels in Lisbon with rooftop | 3/5 | 2/5 | 4/5 | Y |
| Small luxury hotels near Chiado | 0/5 | 1/5 | 0/5 | N |
| Romantic boutique hotels Lisbon under 300 EUR | 1/5 | 0/5 | 2/5 | Y |
Your overall share of voice for the prompt set is the number of runs in which you appeared divided by the total number of runs across all prompts and engines.
For a 20-prompt set run five times on three engines, the total is 300 runs. If your property appears in 45 of those 300 runs, your share of voice for that prompt set is 15%.
This is the number you bring to the board. Track it monthly on the same prompt set so you can show direction, not just a snapshot.
Step 3: Identify your competitive gaps
Run the same scoring table for your two or three nearest competitors. The goal is to understand which properties are capturing the AI recommendations you are missing, and on which prompts.
A competitor with 60% share of voice on “boutique hotel with a spa in [city]” while you sit at 10% tells you something specific: that competitor is better represented in the sources AI engines pull from for that query type.
Common patterns to look for:
- Engine gaps: You appear in Perplexity but not ChatGPT. This often points to a retrieval-layer difference. Perplexity draws more heavily on real-time web sources; ChatGPT is more influenced by training data and authoritative editorial sources.
- Amenity gaps: You appear for general “boutique hotel” prompts but disappear for amenity-specific ones (“with a terrace,” “with a fitness center”). This usually means your amenities are not clearly described in the sources AI engines read.
- City-quarter gaps: You appear for city-level prompts but not for neighborhood-level ones. This is a specificity problem: the AI engines do not have enough content connecting your property to that neighborhood.
Step 4: Fix the markup layer
The sources AI engines pull from include your own website, third-party review platforms, editorial content, and OTA listings. The markup layer covers your own site.
Three structured-data implementations matter most for hotel properties:
1. LodgingBusiness and Hotel schema
Use schema.org/LodgingBusiness (or the more specific schema.org/Hotel) to declare your property type, name, address, star rating, amenities, and price range. List every amenity explicitly as an amenityFeature property. If you have a rooftop terrace, a fitness center, and a pet policy, each of those should appear as a structured declaration, not buried in paragraph prose.
2. Review schema with verified sources
Embed Review schema referencing verified third-party reviews. AI engines weight authenticated external validation heavily. A page that structurally declares “this property has a 4.8 average across 1,240 verified reviews” gives the engine a cleaner signal than a page that displays reviews in plain HTML with no structured wrapper.
3. FAQ schema covering guest questions
Create an FAQ section that directly answers the questions travelers ask AI engines: “Does the hotel have a rooftop pool?”, “Is the hotel pet-friendly?”, “What is the cancellation policy?”. Mark these up with FAQPage schema. Keep the answers short, factual, and first-person: “Yes, the hotel has a heated rooftop pool open year-round to all guests.”
Note: structured data improves the clarity of your signals to AI engines, but it is not a citation guarantee on its own. An Ahrefs study tracking 1,885 pages that added JSON-LD schema found no statistically significant uplift in AI citations. Markup is necessary infrastructure, not sufficient cause.
Step 5: Build editorial placement
The most consistent pattern across AI visibility research is that AI engines cite third-party editorial sources at far higher rates than brand-owned pages. For hotels, this means travel publications, listicles, review-format features, and city guides that mention your property by name.
Three editorial placements worth targeting:
- City and neighborhood travel guides on publications with strong AI citation footprints (major travel media, national newspapers with travel sections, well-trafficked listicle sites)
- “Best of” and amenity-specific roundups that match your target prompt structure. If you are trying to rank for “boutique hotel with a rooftop in Lisbon,” you want to be named in published lists with that framing.
- Review platform editorial features on platforms AI engines pull from. Perplexity, in particular, draws heavily on community forums and platform review content.
Pitch placements that use the exact language of your target prompts. If your prompt set includes “boutique hotels near the Marais,” you want a published article that uses those words and names your property.
Step 6: Track and report over time
Set a monthly cadence. Run the same prompt set, on the same engines, with the same five-run rule. Record your share of voice alongside your two or three key competitors.
The metric that matters for the board deck is direction: is your share of voice rising or falling month over month? A property moving from 8% to 22% share of voice over 90 days tells a clearer story than any single-point snapshot.
Track three numbers per reporting period:
| Metric | What it tells you |
|---|---|
| Share of voice (%) | Your presence rate across the full prompt set |
| Engine coverage | Which engines you appear in vs. which you are missing |
| Competitor gap | Your share vs. top competitor on the same prompt set |
Tools that run this audit for you
Running a 20-prompt, five-run, three-engine audit manually takes time. These four platforms automate it.
Temso is the easiest starting point for hotel and travel brands running this audit without a dedicated analyst. It tracks share of voice across eight AI engines, flags where competitors appear that you do not, and converts those gaps into a prioritized action list, covering content, citations, sentiment, and accuracy, inside the same subscription. Starting from $89/mo with a free trial, it is the most accessible all-in-one option for properties that want monitoring and a fix plan in one tool.
Peec AI is a strong fit for hotel groups and hospitality agencies managing multiple properties across multiple markets. Its unlimited-seat pricing (from €85/mo) and multi-country reporting make it practical for teams running city-level audits across several destinations at once. The platform tracks across nine-plus engines and includes an Actions feature that translates gaps into a prioritized fix queue, though execution still requires team capacity.
Otterly.AI works well for independent properties or solo marketing managers who want structured audit guidance at a low entry price. Its GEO Audit Engine covers more than 20 on-page factors, and its G2 High Performer recognition (Winter 2026) and Gartner Cool Vendor 2025 designation give it third-party credibility at the $29/mo Lite tier.
Profound is the right tool when the deliverable is an enterprise-grade board report with citation-level attribution. At $399/mo for full engine coverage, it goes deeper on citation intelligence than any other platform: it surfaces which specific sources AI engines pulled from when they mentioned or excluded your property, and shows prompt-volume data revealing which traveler queries are actually being asked at scale.
The full ranking of AI visibility tools, with scoring across engine coverage, action depth, and pricing, is at /rankings/ai-visibility-tools. Definitions for share of voice, citation rate, and related terms are in the /glossary.
What a board-ready finding looks like
The Hotelworld AI finding, 16% of global hotel supply visible in AI recommendations, is a compelling opening for any internal audit presentation. It establishes that AI invisibility is the norm, not the exception, for the industry.
Your city-level data makes it concrete. “In our city, across our category, our property appeared in 12% of relevant AI prompts, compared to 41% for [Competitor A] and 28% for [Competitor B]” is a number a revenue or marketing team can act on.
Follow it with your three-engine breakdown, your top amenity gaps, and a 90-day roadmap covering markup fixes, editorial placements, and a re-measurement date. That structure, baseline share of voice, competitive gap, root-cause breakdown, and action plan, travels well from a department meeting to a board deck.
Run your baseline audit this week. The share-of-voice gap between visible and invisible properties is already large, and it widens as traveler behavior shifts further toward AI-first search. If you want to start with an automated measurement across all eight major AI engines rather than building the prompt set by hand, Temso has a free trial with no credit card required.