If 40 percent of U.S. travelers now use generative AI for trip planning, and that cohort skews wealthier, younger, and more college-educated by three to one, what does this mean for the roughly half of American consumers who are not on these rails at all? The hospitality industry is quietly stratifying along an algorithmic axis. This essay asks what that stratification looks like, whether it is stable, and what it means for operators whose economic base is not the AI-native consumer.
By Dan MuirheadResearch by AIVO Agent TeamMay 2026~19 min read
~27%
U.S. adults who have never used
any generative AI platform
63%
Independent hotel bookings
coming through OTAs
~35–40%
Mass-market hotel share of
total U.S. lodging revenue
49→17%
Non-college Americans reporting
6+ close friends, 1990 vs. 2024
From the Founder
What about the other half?
Dan Muirhead · Co-Founder, Head of Strategy
The first two essays in this volume made a specific argument. Essay 01 established that the “return to IRL” narrative describes the social life of roughly the top 10 to 20 percent of American households, and that the affluent cohort driving it is discovering those experiences through AI platforms at roughly three times the rate of non-college-educated Americans. Essay 02 established what it takes for a luxury property to survive AI summarization with its narrative intact: entity authority, citation density, structured data, and editorial presence across the sources AI platforms actually trust.
This essay asks the question underneath both: what about the other half?
AIVO’s work is explicitly in the AI discovery layer. We help brands get found, recommended, and chosen when customers ask AI for answers. That is a real and growing market. But the honest observation, the one most people in AI visibility consulting would rather not say out loud, is that the AI discovery layer is not universal. It is class-sorted. It is age-sorted. It is education-sorted. The roughly 27 percent of American adults who have never used a generative AI platform are not randomly distributed across the population. They are concentrated among the older, the less educated, the lower-income, and the rural. And they are disproportionately the customers of mass-market and midscale hospitality.
This essay is about what that means. For the industry. For operators across segments. And, honestly, for how AIVO thinks about its own addressable market.
01 — Who Is on AI Rails and Who Is Not
The three-to-one gap from Essay 01 extends to every AI platform and to usage depth, not just ever-used
Essay 01 introduced the core finding from Pew’s June 2025 research: 52 percent of Americans with a postgraduate degree have used ChatGPT, versus 18 percent of those with a high school degree or less. The gap in work use is comparable: 45 percent of postgrads versus 17 percent of high-school-or-less. The data has not softened since.
Pew’s September 2025 update on AI in Americans’ lives shows the depth dimension. Among adults with a postgraduate degree, 46 percent interact with AI several times a day. Among those with a high school diploma or less, 20 percent do. The gap is not merely in whether someone has tried ChatGPT once. It is in whether AI is part of the texture of daily life. The awareness gap precedes the usage gap: 60 percent of postgrads have “heard a lot” about AI, versus 38 percent of those without a college education. You cannot adopt a tool you do not know exists.
AI Platform Usage by Education Level
Postgraduate degree
52%Daily: 46%
Bachelor’s degree
51%Daily: ~35%
Some college
33%Daily: ~22%
High school or less
18%Daily: 20%
Source: Pew Research Center (June 2025, n=5,123; September 2025)
Age compounds the gap. Among adults under 30, 58 percent have used ChatGPT and 33 percent interact with AI daily. Among adults 65 and older, 10 percent have used ChatGPT and 5 percent interact daily. That is a fivefold difference. AARP reports that AI usage among adults 50 and older nearly doubled from 18 percent to 30 percent between 2024 and 2025, but from a low baseline that still leaves the majority of older Americans outside the user population entirely.
Geography adds a third axis. Brookings Institution research on AI workforce exposure shows metropolitan knowledge hubs at roughly 40 percent exposure versus 30 percent in rural counties. San Jose, San Francisco, and Washington, D.C. lead. Las Vegas, Toledo, and Fort Wayne lag. The 5-point gap between highly urban and rural counties maps onto the same educational and income stratification. The places with the highest AI exposure are the places with the highest concentration of college-educated, higher-income workers. The places with the lowest are the places that also lost their churches, their union halls, and their VFW posts.
ChatGPT Adoption by Age
18–29
58%Work: 38%
30–49
41%Work: 30%
50–64
25%Work: 18%
65+
10%Work: n/a
Source: Pew Research Center (June 2025)
The aggregate number: roughly 27 percent of U.S. adults have never used any generative AI platform, down from 33 percent six months earlier. That is progress. It is also still one in four Americans who are entirely outside the user population for these tools. And that one in four is not randomly distributed. It is older. It is less educated. It is lower-income. It is rural. It is, disproportionately, the customer base of economy and midscale hospitality.
Now layer this on travel. Phocuswright’s November 2025 report found that nearly 40 percent of U.S. travelers used generative AI for trip research in the previous 12 months, an 11-point increase in a single year. The profile of the AI-using traveler is by now familiar: younger, higher household income, more trips per year including international, higher spend per trip. McKinsey via OAG puts it higher: more than half of travelers use ChatGPT or equivalent at least occasionally for trip planning, with only 11 percent refusing to use AI at all.
The interesting wrinkle is that comfort with AI for travel is declining among the youngest cohort. YouGov data shows comfort dropping from 47 percent to 34 percent among 18-to-24-year-old travelers between 2024 and 2025, a 13-point decline. Meanwhile, comfort ticked up among Americans 55 and older, from 16 percent to 20 percent. The adoption curve is not a clean line. But the class gradient underneath it is consistent.
“AI-using travelers are a high-value segment. They skew younger yet report higher median household income, take more trips including international, and spend more annually on travel.”
02 — What the Mass-Market Discovery Layer Actually Looks Like
Google search, OTAs, loyalty programs, family referral, road trip habit
The roughly half of American travelers who are not using generative AI for trip planning are not wandering in the dark. They have a discovery layer. It is just not the one the AI visibility industry talks about.
For independent hotels, which comprise a significant share of economy and lower-midscale inventory, online travel agencies accounted for 63.4 percent of bookings in 2025, up from 61.3 percent in 2024. The trend is accelerating, not moderating. These operators are shedding direct bookings and becoming more platform-dependent with each passing quarter. The true cost of OTA distribution, when you account for 15 to 20 percent commission rates, 50 percent cancellation rates on Booking.com (versus 18 percent on direct channels), lost guest data, and reduced upsell opportunity, runs to roughly 30 to 35 percent of booking revenue.
Booking Channel Mix: Mass-Market Independent vs. Luxury Branded
Google search remains the mass-market hotel industry’s primary paid discovery mechanism beyond OTAs. The hospitality cost-per-click on Google Ads reached $1.63 in Q1 2026, up 7 percent year over year. For a property earning $47 RevPAR in the midscale-economy tier, the acquisition math on paid search is punishing. A $150 room night booked through Booking.com at 20 percent commission nets the hotel $120. The same room booked through Google Ads at a $100 customer acquisition cost nets $50. Mass-market operators often find that direct bookings acquired through expensive paid channels yield less net revenue than OTA bookings despite the absence of commission.
Loyalty programs are the third pillar. Marriott Bonvoy alone has 271 million members, and member stays account for 75 percent of room nights in the United States and Canada. Hilton Honors follows at roughly 210 million. Wyndham Rewards at 114 million. Choice Privileges at 69 million. Total membership across major brands reached 675 million in 2024, growing 14.5 percent year over year, more than double the 6.7 percent growth in room supply. Repeat loyalty guests spend 22.4 percent more and stay 28 percent longer.
But loyalty economics work differently by segment. At the luxury tier, loyalty members tend to be genuine frequent travelers and elite status holders. At the economy tier, loyalty membership has been inflated by credit card partnerships and airline co-marketing, and the member quality often reflects discount-seeking behavior rather than brand preference. The cost of loyalty program participation, $5.46 per occupied room and rising faster than revenue, pressures mass-market properties where margins are thinnest.
The fourth channel, and the one least discussed in any industry report, is family referral, repeat habit, and word of mouth. The truck driver who stays at the same Motel 6 on the I-40 corridor twice a month. The family that books the same Days Inn in Pigeon Forge every July. The road-trip traveler who searches “hotels near me” on Google Maps at 9 p.m. and picks whatever is cheapest within two miles. These patterns do not show up in Phocuswright surveys or AI adoption studies, but they account for a substantial share of economy and midscale occupancy. The discovery layer for this traveler is not ChatGPT. It is the highway sign, the Google Maps pin, the loyalty card in the wallet, and the recommendation from a brother-in-law.
03 — The Squeezed Middle and the Bifurcation
What is happening structurally as the top end goes AI-native and the bottom end stays in the older channel mix
The U.S. hotel industry’s performance data in 2025 tells a K-shaped story. Luxury properties recorded RevPAR growth of approximately 3 percent, driven entirely by rate increases. Economy properties recorded demand declines exceeding 3 percent, ADR erosion surpassing 2 percent, and RevPAR contraction of 4.4 percent. Economy properties missed their budget by 12.8 percent. Luxury missed by 7.2 percent. The spread between the two is not cyclical. It is structural.
The K-Shaped Hotel Industry, 2025–2026
Luxury & upper upscale1.3M (22.5%)$186~50%
Upscale & upper midscale2.5M (42.6%)$99~35%
Midscale & economy2.0M (34.9%)$47~15–20%
Source: MMCG Invest / STR / CBRE (2026)
The revenue concentration is stark. Luxury and upper upscale properties hold 22.5 percent of U.S. room inventory and generate roughly 50 percent of total hotel revenue. Midscale and economy properties hold 34.9 percent of inventory and generate roughly 15 to 20 percent of revenue. The 2 million rooms in the midscale-economy tier, collectively, produce less revenue than the 1.3 million rooms in the luxury tier. The RevPAR gap ($186 versus $47) is a fourfold difference, and it is widening.
The upper-midscale segment, the largest by both existing rooms and construction pipeline, occupies the middle ground. It recorded only a 3.3 percent shortfall against budget, far better than economy’s 12.8 percent, and its select-service model delivers 40 to 50 percent gross operating profit margins, well above the 25 to 35 percent typical of full-service properties. Capital is flowing here: upper midscale accounts for 30 percent of all rooms under construction. But this segment faces its own squeeze.
From below, Airbnb reported 13 percent revenue growth in 2025 and 1.76 million active U.S. listings, up 6.1 percent year over year. For the middle-income leisure traveler who historically anchored upper-midscale occupancy, Airbnb’s supply depth and pricing flexibility increasingly compete directly. From above, Marriott Bonvoy’s 271 million members create a gravitational pull toward branded properties that independent operators cannot replicate. A midscale independent hotel has no equivalent mechanism to capture the psychology of accumulated loyalty points and elite benefits.
Hotel REIT and private-equity capital has recognized this bifurcation. Investment is flowing to upscale select-service properties with defensible margins and away from true economy, where compressed margins and limited pricing power no longer justify the capital. The construction pipeline confirms it: 75 percent of rooms under construction are limited-service, and economy-tier development has contracted in favor of extended-stay and upper-midscale formats that can still generate acceptable returns.
The AI discovery layer adds a new axis to this bifurcation. Luxury brands are embedding their properties into AI systems through API integrations, comprehensive schema markup, and commercial agreements that ensure their inventory surfaces in AI-powered recommendation engines. Mass-market properties, particularly independents, often operate with legacy property management systems, minimal data infrastructure, and no capability to implement or maintain structured data. BCG’s 2026 analysis is direct: hotels that invest in data integration and AI-native distribution will own the guest relationship and margin. Hotels that do not may not appear at all if the AI cannot see them.
The result is a vicious cycle. Mass-market properties generate lower visibility in AI discovery, acquire fewer bookings through algorithmic channels, achieve lower occupancy and RevPAR, and therefore lack the financial resources to invest in the data infrastructure required to compete. The properties that need AI visibility most are the properties least able to build it.
04 — The Larger Social Dimension
What it means that community infrastructure has collapsed for one half of the country while the other half pays premium prices for engineered versions of it
This is the part of the argument the volume has been building toward since Essay 01, and it deserves to be said plainly.
Over the last thirty years, the institutional infrastructure of American community life has collapsed for non-college-educated Americans. The data is not ambiguous.
Regular religious attendance has fallen from 42 percent to 30 percent of the adult population over two decades. Church membership has dropped below 50 percent for the first time in Gallup’s polling history. Union membership has been cut in half, from 20.1 percent in 1983 to 10.0 percent in 2025. In the private sector, where working-class Americans are concentrated, it is 5.9 percent. The VFW has lost roughly half its membership over thirty years. Masonic membership is down approximately 75 percent from peak. The Elks, the Knights of Columbus, the American Legion, bowling leagues, the whole apparatus Robert Putnam documented in Bowling Alone, all of it has contracted severely.
The Institutional Collapse
Regular church attendance42% of adults30% of adults-28%
Union membership (all)20.1% (1983)10.0% (2025)-50%
Union membership (private sector)~25% (1983)5.9% (2025)-76%
VFW membership~3M (peak)~1.5M (2025)-50%
Masonic membershipPeak level~25% of peak-75%
Source: Gallup (2024); BLS (Jan 2025); Stars and Stripes (2025); Davis Lodge/industry data
The AEI Survey Center’s 2024 study provides the number that anchors this volume: 49 percent of Americans with a high school degree or less reported having six or more close friends in 1990. By 2024, that number was 17 percent. Twenty-four percent of non-college-educated Americans now report having no close friends at all. One in five Americans lives in a community with no access to any public or commercial gathering space.
For working-class Americans, the places where social life happened, the church, the union hall, the VFW post, the bowling league, the diner, the local bar, have been substantially dismantled. The casual dining chains that served as residual third places are themselves contracting: Denny’s closed 88 restaurants in 2024. The independent restaurant sector shrank 2.3 percent in 2025. In small-town and rural America, the infrastructure of social connection is not being replaced. It is simply disappearing.
Meanwhile, the affluent cohort that can afford to pay for connection has built a commercial version of it. Soho House’s 271,500 members. Zero Bond’s $5,000 initiation. Aman’s $200,000 New York membership. The run clubs, the supper clubs, the phone-free retreats, the $7,000 curated villa in Oaxaca. These are real products serving real needs. They are also luxury goods in the classical sense, available to the Americans whose community infrastructure never collapsed because they could always afford to rebuild it.
The AI discovery layer sits on top of this stratification and reinforces it. The affluent, college-educated, under-45 traveler who uses ChatGPT to find a phone-free dinner in Austin is the same person who can afford the dinner and who lives in a social world where someone mentioned it. The non-college-educated American in a town where the VFW post closed and the Applebee’s shut down is not asking ChatGPT for restaurant recommendations. The question is not whether they could. It is whether the channel is relevant to the life they are actually living.
“The 49-to-17-percent collapse in close friendships among Americans with a high school degree or less is the biggest social story of the last thirty years. It is not the story the industry is telling.”
05 — So What
Three audiences, three honest answers
For operators targeting the AI-native affluent traveler: The first two essays in this volume laid out the playbook. AI visibility is how your highest-value customer finds you. Entity authority, citation density, structured data, and editorial presence across the sources AI platforms trust are now operating requirements, not marketing nice-to-haves. The 40 percent penetration rate is rising 11 points a year. The brands that build their AI infrastructure now will own this channel. The brands that wait will spend years trying to recover visibility in a system that compounds early investment. This is the AIVO thesis, and the data supports it.
For operators targeting mass-market and midscale travelers: AI visibility is not your top-five strategic priority this year, and anyone who tells you otherwise is selling something. Your customers are finding you through OTAs, Google search, loyalty programs, and word of mouth. Your economic challenge is margin compression from OTA commissions, rising Google CPC costs, and Airbnb supply competition. Your strategic question is whether to invest in direct booking infrastructure, loyalty program optimization, and rate management, or whether to accept OTA dependency and optimize within it. The AI discovery layer is relevant to approximately 15 to 25 percent of your customer base right now, weighted toward the highest-spend segment. It will become more relevant over the next five years as adoption broadens. But it is not where your bookings come from today, and pretending it is will lead to misallocated capital.
01
Luxury & Upper Upscale
First-order strategic priority
Build entity authority, citation density, and structured data now. The 40% AI traveler penetration rate is concentrated in your customer base. Every quarter of delay is compounding lost ground.
02
Mass-Market & Midscale
Secondary priority
Your primary challenges are OTA margin compression, Google CPC inflation, and Airbnb competition. Invest in direct booking infrastructure and loyalty optimization first. Monitor AI adoption in your customer base and build readiness, not urgency.
03
Investors & Strategists
Structural, not cyclical
The K-shaped bifurcation is structural, not cyclical. Capital allocation should reflect the AI-mediated discovery advantage that luxury and upper-upscale properties increasingly hold. Mass-market properties without data infrastructure face a compounding visibility disadvantage that will not self-correct.
For investors and strategists thinking about the next decade: The K-shaped bifurcation in the hotel industry is structural. Luxury and upper upscale hold 22.5 percent of rooms and generate roughly 50 percent of revenue. The RevPAR gap is fourfold and widening. AI-mediated discovery adds a new stratum: properties with data infrastructure and AI visibility will capture disproportionate share of the highest-value bookings, while properties without it will become algorithmically invisible to the most lucrative traveler segment. The construction pipeline is already reflecting this, with capital flowing to upscale select-service and away from economy. The question for the next decade is whether the midscale “squeezed middle,” caught between Airbnb and Marriott Bonvoy, can find a viable economic position, or whether the industry continues to bifurcate until the middle tier contracts.
Conclusion
The right posture is calibration, not pretense
AIVO’s work is in the AI discovery layer. This is not a neutral position. We believe that AI-mediated discovery is reshaping how affluent travelers find, evaluate, and book hospitality experiences, and that brands which invest in visibility across this layer will capture disproportionate value. The data supports that belief, and it supports it more strongly with every quarter.
But the AI discovery layer is not universal. It is class-sorted, age-sorted, education-sorted, and geography-sorted. The roughly 27 percent of American adults who have never used a generative AI platform, and the larger share who use it rarely or not for travel, are not a rounding error. They are half the country. They are the customers of 2 million midscale and economy hotel rooms. They are the Americans whose churches closed, whose union halls disappeared, whose VFW posts shut down, and whose diners are contracting. Their discovery layer is Google Maps, Booking.com, the Wyndham Rewards card, and the recommendation from a family member. That layer is not broken. It is not obsolete. It is not waiting for AI to replace it. It is the actual channel through which mass-market hospitality operates.
The industry stratification this volume describes is real. The top end of American hospitality is going AI-native in its discovery layer, and the properties that invest in this transition will own the highest-value customer segment. The bottom end is staying in the older channel mix, and the operators who serve it well will continue to do so through OTAs, loyalty programs, and direct relationships. The operators who get hurt are the ones in the middle who misread which layer their business actually lives in.
The right posture for any operator is calibration. Know which customer you have. Know which discovery layer they use. Invest accordingly. Do not pretend that AI visibility is equally urgent for every segment, and do not pretend it is irrelevant for any of them. The honest version of the story is more useful than the salesy one, and the industry deserves the honest version.
The volume ends here. Three essays, one argument: the experience economy is real, the AI discovery layer underneath it is real, and both are class-sorted in ways the industry has not yet come to terms with. The operators who understand the shape of what is happening, and who know which part of it they are operating inside, are the ones who will build correctly for the next decade. The ones who mistake the trend-deck version for the whole story will build for a customer who does not exist in the numbers the language implies.
Sources and Further Reading
Pew Research Center, U.S. Adults and ChatGPT (June 2025)
Pew Research Center, AI in Americans’ Lives: Awareness, Experiences, and Attitudes (September 2025)
Pew Research Center, Teens, Social Media and AI Chatbots (December 2025)
Pew Research Center, Demographic Differences in How Teens Use and View AI (February 2026)
Gallup, Gen Z AI Adoption Steady, Skepticism Climbs (2026)
Gallup, Church Attendance Has Declined Across Religious Groups (2024)
AARP, 2026 Technology Trends: Older Adults (2026)
Quinnipiac University Poll, AI Usage Survey (October 2025)
Brookings Institution, The Geography of Generative AI’s Workforce Impacts (2025)
McKinsey, Gen AI and the Racial Wealth Gap (2025)
YouGov, Comfort with AI in Travel Planning Dips Among Younger Americans (2025)
Phocuswright, Search Slips, AI Surges: Travel’s New Front Door? (November 2025)
McKinsey via OAG, Travel 2045: A 20-Year Outlook for the AI Era (December 2025)
MMCG Invest / STR / CBRE, U.S. Hotel Industry Analysis (2026)
HITEC, A Two-Speed Industry: How Bifurcation Reshaped U.S. Hotel Performance in 2025 (2026)
Phocuswire / Cloudbeds, Hotel Distribution and AI Trends (2025)
Marriott International, Fourth Quarter and Full Year 2025 Results
OysterLink, Hotel Loyalty Program Statistics (2024–2025)
Bureau of Labor Statistics, Union Members Summary (January 2025)
Cox, D. and Pressler, S., Disconnected: The Growing Class Divide in American Civic Life, AEI Survey Center on American Life (August 2024)
Stars and Stripes, Veterans Organizations Membership Decline and Relevance (October 2025)
Putnam, R., Bowling Alone (2000) and The Upswing (2020)
NRN, The Independent Restaurant Sector Shrunk by 2.3% in 2025 (2025)
