What research from Accenture, Bain, BCG, Deloitte, McKinsey, PwC, and Skift means for hotel discovery, accurate representation, and commercial readiness.
A hotel can appear in an AI recommendation and still be described incorrectly. An answer that misstates beach access or omits a relevant room configuration can mislead a traveler even when it includes the hotel and a working booking link.
For hotel leaders, being found, being understood, and being bookable require separate attention. Recent consulting research shows why these distinctions matter as AI enters travel planning. AGR’s contribution is to apply Knowledge Formation Optimization, or KFO, to the public information behind the hotel’s representation and to the evidence needed to assess it.
AI-assisted research is already consequential
Deloitte’s 2026 Summer Travel Survey reports that 25% of its traveler subgroup used generative AI for travel research. The planning question covered 1,808 summer travelers within an April 2026 survey of 4,003 Americans. A separate question covering 446 AI travel researchers shows several types of tools in use, including standalone assistants, search engines, and travel sites. Deloitte, 2026 Summer Travel Survey, methodology and page 29.
Accenture’s June 2026 survey of 5,003 leisure travelers across 12 markets shows how readiness to delegate varies by task: 40% said they would fully delegate comparison of travel options to an AI agent, while 10% would fully delegate payment execution. These are expressions of willingness, not observed booking behavior. Accenture, “Win the AI agent, win the traveler,” Figure 2 and methodology.
Skift’s State of Travel 2025 reproduces findings from its U.S. Traveler Trends 2025 survey (n=1,002; data as of June 2025). It reports that 46% preferred to make bookings themselves using AI suggestions and 41% preferred AI booking with prior confirmation. Just 2% selected full autonomy to book or modify arrangements after initial guidance. These are preferences for control, not measured booking adoption. It also reports high trust in AI travel information among 93% of respondents. Skift, State of Travel 2025, section 3.6, page 89 (PDF hosted by Tellusant); originating report: U.S. Traveler Trends 2025 (subscription or purchase).
For AGR, the implication is to prepare for AI-assisted comparison while travelers retain booking authority. The assistant’s account of the hotel can influence consideration even when it never handles the reservation.
The eventual distribution structure remains uncertain. McKinsey describes four possible futures: copilot commerce, AI-powered experience curation, a direct-booking renaissance, and agent takeover. These scenarios allow different roles for hotel websites, online travel agencies, and AI intermediaries. McKinsey, “How could hotel booking be disrupted by agentic AI?”.
Hotels can improve their representation across the environments relevant to their guests without making the strategy depend on one forecast prevailing.
Being named and being understood are different outcomes
Counting appearances measures whether the hotel is present. Assessing the answer’s content reveals whether it gives the traveler a sound basis for choosing.
Skift’s survey findings include reports of generic or non-personalized suggestions from 46% of respondents and outdated or incorrect information from 39%. These self-reported experiences concern AI travel tools generally; they do not establish a hotel-specific error rate. Skift, State of Travel 2025, section 3.6, page 90 (PDF hosted by Tellusant).
Consider a hypothetical resort with connecting family suites and beach access by scheduled boat. An AI answer might include it in a family shortlist but describe it as having a beach directly outside the rooms. Another answer might describe the transfer correctly but omit the connecting-bedroom configuration the family needs.
The first answer contains a factual error. The second loses a relevant distinction. An appearance count would miss both problems.
Relevance depends on the traveler. Drawing on 2025 U.S. travel surveys, Deloitte’s 2026 Travel Industry Outlook describes different priorities among generations of luxury travelers, including location, food, family experiences, and room or spa amenities. These findings concern guest preferences, rather than AI performance, but they help explain why a generic luxury description can be inadequate. Deloitte, 2026 Travel Industry Outlook.
AGR’s interpretation is that the property record should make these distinctions explicit: what the hotel offers, whom particular accommodations suit, what qualifications apply, and which claims can be substantiated. Accuracy includes preserving meaningful limits. The account should help the traveler judge the property as it actually is.
The relevant information extends beyond the hotel website
BCG recommends broad, detailed, machine-readable hotel content supported across multiple sources. Its companion hospitality analysis also recommends checking whether online representation matches the experience guests receive. BCG, “AI-First Hotels”; BCG, AI Disruption in Hospitality, printed page 17.
Bain’s analysis with Meikai provides adjacent evidence from luxury personal goods. It finds substantial third-party citation in unbranded AI responses and recommends external-source work as part of generative engine optimization, or GEO. Hospitality was explicitly excluded, so its citation findings should not be transferred to hotels. Bain, “Winning Over the Customer in the Age of AI,” including methodology.
McKinsey’s cross-industry AI-search analysis includes hotels among categories where leading brands can be absent from some answers. This is an observation in its broader analysis, rather than a hotel-specific prevalence estimate. It also describes source composition as varying with the platform, location, category, and question. The analysis supports testing actual answers rather than assuming that established brand strength ensures inclusion. McKinsey, “New front door to the internet”.
Both Bain and McKinsey include external content and continuing measurement within GEO. KFO shares some of that work and organizes responsibility for the public source record: how the property is defined, how its claims are supported and corrected, and whether AI answers preserve those distinctions across relevant questions and over time.
What KFO makes the program accountable for
Americas Great Resorts defines Knowledge Formation Optimization as follows:
KFO structures, sequences, distributes, corroborates, and corrects intellectual frameworks and entity definitions across the public information environment and measures whether AI systems reproduce them accurately across relevant queries and over time.
Americas Great Resorts, canonical KFO framework.
AGR recommends that a hotel applying KFO begin with a dated factual baseline identifying the authoritative basis for each material claim. That record should establish the property’s identity, location, accommodations, amenities, restrictions, and substantiated distinctions. It should separate current facts from historical descriptions and independent corroboration from multiple copies of the same supplied text.
The next step is to compare that record with public sources and AI answers to relevant guest questions. If an answer confuses two similarly named hotels, inspect the names, locations, and relationships established in the source material. If it repeats an obsolete amenity, identify where the outdated description remains and seek corrections supported by current evidence. Record each change and check whether subsequent answers reproduce the facts more accurately.
The practical output is a maintained property record, a documented correction history, and repeated observations that assess inclusion separately from factual accuracy and relevant omissions. This connects source work to an explicit standard for evaluating results.
This approach is AGR’s interpretation of the consulting research, which does not test KFO or establish its superiority. Citations do not imply endorsement by the firms named. Hotel teams can improve information they control, seek substantiated corrections elsewhere, and measure public answers. Those observations do not expose hidden model decisions or guarantee a particular recommendation.
Accurate representation and booking readiness must work together
A traveler needs a reliable explanation of the experience. An agent executing a booking also needs current availability, prices, conditions, and an authorized way to act.
McKinsey and Skift’s “Remapping travel with agentic AI” distinguishes advisory assistance from systems able to carry out tasks. It describes agents using both structured information and contextual material such as reviews and travel writing. The same journey can therefore require coordinated content and technical work. McKinsey and Skift, “Remapping travel with agentic AI”.
PwC recommends structured content, inventory and API readiness, loyalty integration, and continuing visibility monitoring. PwC, “The future of agentic commerce in travel is taking off”.
Accenture makes the discovery implication explicit: hotels that fail to publish availability, policies, and loyalty rules in agent-readable form risk exclusion from AI-native recommendations before a transaction begins. This is strategic advice about agent-mediated travel, not a measured causal relationship between booking connectivity and general public AI recommendations. It strengthens the case for coordination while leaving those outcomes to be tested separately. Accenture, hotel recommendations, page 19.
A hotel-owned assistant, an assistant inside an online travel agency, and a general public AI answer are different environments. Record results in the environment where they occur. A successful reservation test demonstrates execution; a public-answer review assesses how the property is presented to prospective guests.
Measure the outcomes that matter to guest decisions
AGR recommends testing questions that name the hotel, questions about relevant needs without naming it, and comparisons where its characteristics matter. A family holiday, a quiet couples’ stay, and a meeting requiring specific facilities call for different evidence of suitability.
AGR recommends assessing the following outcomes separately:
| Question | Evidence to retain |
|---|---|
| Is the hotel included in relevant answers? | A defined set of queries, repeated observations, and the proportion containing the property. |
| Is the hotel described accurately and usefully? | Material claims, relevant omissions, and comparison with the dated factual baseline. |
| Which sources are visible? | Displayed citations and links, source dates, and identifiable contradictions. Citations reveal only the sources shown. |
| Can a supported booking journey execute correctly? | Separate tests of availability, prices, conditions, booking, and servicing in the relevant environment. |
| Is there a commercial effect? | Identifiable referrals, inquiries, bookings, and guest feedback, with attribution gaps recorded. |
For AI-answer observations, retain the date, platform, available model or mode information, exact question, session conditions, full response, and visible citations. Use a consistent protocol across relevant questions and over time, retaining unfavorable results alongside favorable ones.
Record source changes alongside those observations. Better answers following a correction establish an association; platforms, competing sources, and retrieval behavior can also change. Commercial impact requires separate evidence connecting the work to inquiries, bookings, or revenue.
This work needs an accountable owner and cooperation among property operations, marketing, distribution, and technology. Operations confirms what guests receive; content teams maintain the description; distribution and technology maintain executable offers. McKinsey’s broader organizational analysis reinforces the management case for workflow redesign, governance, and human judgment. McKinsey, “AI is everywhere. The agentic organization isn’t yet”.
The leadership standard is clear: a prospective guest should encounter an account of the hotel that is accurate, relevant to the question, and supported by current evidence. KFO organizes responsibility for that source record and measures how AI answers reproduce it. For hotel leaders, the commercial objective is to preserve the facts and distinctions on which an informed guest decision depends.
