A 240-prompt study across ChatGPT and Perplexity shows the 'Big Six' capture 78% of budget queries, but plunge to 39% when travelers ask for the 'best' hotel.
Large hotel chains dominate AI recommendations when travelers specify a budget, but lose ground when users ask for the best places to stay.
A study conducted on October 29, 2026, by 5W Public Relations analyzed 240 hotel recommendations returned by ChatGPT (gpt-5.5) and Perplexity (sonar-pro) across six major U.S. markets: New York City, Chicago, Nashville, Miami, Savannah, and San Diego.
The findings show that hospitality brands face a bifurcated search ecosystem: luxury and quality queries run on property-level editorial prestige, while budget queries run on structured pricing and directory inventory.
The AI Visibility Split: Price Versus Prestige
Across 24 test prompts, the "Big Six" hotel operators—Marriott International, Hilton, IHG, Hyatt, Wyndham Hotels & Resorts, and Choice Hotels—captured 77.5% of budget recommendations (93 of 120 picks). When prompted for the "best hotels to stay at" without pricing limits, that share dropped to 39.2% (47 of 120 picks).
The number-one recommendation slot mirrored this gap. Big Six brands claimed 9 of 12 top placements for budget inquiries, but only 2 of 12 top placements for open-ended "best" inquiries.
| Traveler Query Type | Total Hotel Picks | Big Six Picks Won | Big Six Win Rate | Top #1 Placements | Picks With Parent Name |
|---|---|---|---|---|---|
| Best-Hotel Search (Open budget) | 120 | 47 | 39.2% | 2 of 12 | 8 of 47 (17.0%) |
| Budget Search (~$200/night family) | 120 | 93 | 77.5% | 9 of 12 | 39 of 93 (41.9%) |
The geographic disparity was sharpest in New York City. Big Six brands captured 70.0% of NYC budget selections, but secured only 10.0% (2 of 20 picks) on best-hotel lists. Savannah was the only market where major chains captured a majority of best-hotel selections, winning 65.0% (13 of 20 picks).
The Master Brand Disappears in Luxury Citations
When legacy hotel groups win premium AI recommendations, their corporate identity rarely appears.
Only 17.0% (8 of 47) of Big Six luxury recommendations included the corporate parent name in the output. By comparison, 41.9% of budget recommendations carried the parent name.
| Hotel Parent Company | Best-Hotel Wins | Wins With Parent Name | Example Properties Recommended by AI |
|---|---|---|---|
| Marriott | 23 (of 120) | 2 | The Joseph; The Ritz-Carlton, Chicago; W Nashville; The Miami Beach EDITION |
| Hyatt | 11 (of 120) | 5 | Park Hyatt Chicago; Thompson Nashville; Chicago Athletic Association Hotel; Andaz Savannah |
| Hilton | 9 (of 120) | 1 | Waldorf Astoria Chicago; Conrad Nashville; Hotel del Coronado; The Drayton Hotel |
| IHG | 4 (of 120) | 0 | InterContinental Miami; InterContinental San Diego; Kimpton Brice Hotel |
| Wyndham | 0 (of 120) | 0 | None recommended |
| Choice | 0 (of 120) | 0 | None recommended |
Marriott led all parent operators with 23 luxury selections, yet only two of those responses mentioned Marriott directly. Models surface soft brands like Autograph Collection or luxury flags like The Luxury Collection by individual property names, leaving the master brand uncredited in generative answers. This confirms patterns documented in 5W's 50-State Hotel AI Map.
Editorial Press Feeds Luxury Answers; Direct Feeds and OTAs Feed Budget
The inputs retrieved during web searches dictate AI outputs. An analysis of 120 Perplexity source links revealed distinct retrieval pathways for each query type.
| Digital Information Source | Best-Hotel Queries (of 60 links) | Budget Queries (of 60 links) |
|---|---|---|
| Travel Magazine Editors (Condé Nast Traveler, U.S. News, Time Out, Michelin Guide) | 27 (45.0%) | 5 (8.3%) |
| Booking Engines & Review Platforms (Hotels.com, Tripadvisor, Expedia, Kayak) | 19 (31.7%) | 35 (58.3%) |
| Official Chain Direct Websites (choicehotels.com, ihg.com, marriott.com) | 9 (15.0%) | 20 (33.3%) |
| Local Tourism Boards & Brand Content | 5 (8.3%) | 0 (0.0%) |
ChatGPT reflected the same mechanism. When processing quality-focused searches, ChatGPT queried publications by name in 5 of 6 instances (such as "Michelin Guide best hotels Chicago 2026"). For budget searches, the model queried price and occupancy attributes (such as "Chicago family hotels under $200 per night neighborhood").
Citation counts do not guarantee hotel placements. Perplexity cited choicehotels.com across 4 of 6 best-hotel searches because the domain holds directory indexation, yet zero Choice properties appeared in the recommended lists. Chain directory pages confirm inventory but do not supply the qualitative endorsement generative models require for editorial answers. Directing digital investment toward property-level public relations rather than corporate site copy aligns with how models select recommended properties, as detailed in 5W's guide to hotel SEO and PR for direct bookings.
Engine Divergence Across Query Formats
ChatGPT and Perplexity converged on luxury recommendations, sharing 63.3% of their top picks (38 of 60). Because both systems retrieve from a narrow cluster of legacy travel guides, their premium recommendations align.
Budget recommendations diverged sharply. The two models shared 11.7% of budget picks (7 of 60).
| Query Category | Total Picks Per Engine | Identical Picks Shared by Both | Cross-Engine Agreement Rate |
|---|---|---|---|
| Best-Hotel Inquiries | 60 | 38 | 63.3% |
| Budget Inquiries | 60 | 7 | 11.7% |
In New York City, Perplexity highlighted six Hilton properties, while ChatGPT selected two different Hilton hotels, leaving zero overlap between the lists. Because booking directories contain thousands of comparable mid-tier properties, each engine retrieves a different subset. Multi-platform tracking across separate search models remains necessary to evaluate portfolio visibility.
Generative Link Routing Creates Friction for Chain Properties
Auditing the ChatGPT consumer interface for Nashville luxury hotels revealed a technical tracking breakdown for major hotel portfolios. Independent properties and boutique operators received direct links to their specific property pages, whereas all five Big Six hotels routed users to generic corporate homepages.
| Recommended Nashville Property | Parent Hospitality Group | Destination URL Provided in ChatGPT Body Copy |
|---|---|---|
| Four Seasons Hotel Nashville | Four Seasons | fourseasons.com/nashville (Direct Property Link) |
| The Hermitage Hotel | Independent | thehermitagehotel.com (Direct Property Link) |
| JW Marriott Nashville | Marriott | marriott.com (Generic Brand Homepage) |
| The Joseph, a Luxury Collection Hotel | Marriott | marriott.com (Generic Brand Homepage) |
| Noelle | Independent | noelle-nashville.com (Direct Property Link) |
| Thompson Nashville | Hyatt | hyatt.com (Generic Brand Homepage) |
| Omni Nashville Hotel | Omni | omnihotels.com/hotels/nashville (Direct Property Link) |
| The Union Station Nashville Yards | Marriott | marriott.com (Generic Brand Homepage) |
| Conrad Nashville | Hilton | hilton.com (Generic Brand Homepage) |
| The Russell | Independent | therussellnashville.com (Direct Property Link) |
While the interactive map module above the response listed the direct property links, the response copy sent prospective guests to top-level domains. This disconnect inserts extra navigation steps between the search query and the final reservation.
Action Items for Hospitality Brands in Generative Search
| Strategic Marketing Priority | Operational Objective | Verification Method |
|---|---|---|
| Pitch Editorial Outlets for Soft Brands | Secure named features in tier-one lifestyle publications (Condé Nast Traveler, Travel + Leisure). AI models rely on these editorial guides for premium recommendations. | Query Perplexity for top properties across core markets and audit citations for brand inclusion. |
| Establish Narrative Angles for Individual Properties | Provide clear property narratives (architecture, art collections, historic renovations) that editors and AI models can reference. | Prompt ChatGPT to describe property features and benchmark against marketing copy. |
| Standardize Schema Markup for Property URLs | Implement schema.org Hotel structured data across localized pages and press releases to prevent citations from routing to corporate homepages. | Monitor output links within ChatGPT conversational responses across core markets. |
| Maintain Consistency Across Inventory Directories | Ensure pricing, neighborhood tagging, and room specs match across online travel agencies and direct sites to maintain budget visibility. | Audit consistency of rates and local neighborhood descriptions across third-party OTAs. |
| Audit AI Visibility by Platform and Market | Track brand citations across ChatGPT, Perplexity, and Google AI Mode independently to account for source divergence. | Run monthly prompt tracking across engines to isolate platform-specific gaps. |
Early narrative placement determines how generative models index a hotel, as outlined in 5W's hotel opening PR guide. As consumer behavior shifts from static search to conversational discovery, hotels must optimize their technical footprints for AI parsing, an evolution examined in 5W's report on how ChatGPT is changing trip planning.
Research Methodology
This study tested 24 automated search queries on October 29, 2026, using the DataForSEO AI Optimization API with live web retrieval enabled.
Engines Evaluated: OpenAI ChatGPT (gpt-5.5) and Perplexity (sonar-pro).
Markets Evaluated: New York City, Chicago, Nashville, Miami, Savannah, and San Diego.
Query Prompts: Best-Hotel Evaluation ("List the 10 best hotels to stay at in [City, State]. Give only the hotel name and neighborhood for each, as a numbered list."); Budget Evaluation ("List 10 hotels in [City, State] for a family of four with a budget of about $200 a night. Give only the hotel name and neighborhood for each, as a numbered list.").
Dataset Scope: 24 total queries returning 240 property picks and 120 Perplexity source links.
Categorization Rules: Major parent networks include Marriott International, Hilton, IHG, Hyatt, Wyndham Hotels & Resorts, and Choice Hotels. Properties were classified as parent-attributed when the model-generated output explicitly included the corporate parent name. Country Inn & Suites by Radisson was classified under Choice Hotels. Timeshare properties operating under licensing agreements were tracked as separate entities.
