Vertical Market Evolution
Hospitality and Lodging: Where Guest Discovery Is Headed
OTAs still take a growing share of independent-hotel bookings, but the bigger shift is in where travelers start looking, and no one has mapped that terrain for hospitality yet.
Part of Vertical Playbooks in the Insights library.
Abstract
OTAs still capture roughly a quarter of global travel spending, and more strikingly, 63.4 percent of independent-hotel bookings worldwide in 2025, up from 61.3 percent a year earlier, while the cost of winning that guest keeps outpacing revenue. At the same time, the starting point for trip research is moving toward generative AI tools faster than any hospitality-specific measurement system exists to track it: U.S. search engine use for travel research fell from 51 percent to 36 percent in a single year while AI tool use for travel more than doubled. The real question for a lodging brand isn't OTA versus direct anymore, it's where guest attention is actually migrating across search, reviews, social, marketplaces, and the still-unmeasured AI-answer layer.
Hospitality's discovery and booking channels are shifting on two fronts at once, and they're not moving at the same speed. On the transaction side, the balance between OTAs and direct booking has barely budged at the headline level in a decade: Hilton's 2016 'Stop Clicking Around' campaign didn't meaningfully dent OTA distribution share industry-wide, and independent-hotel reliance on OTAs actually rose to 63.4% of bookings globally in 2025, up from 61.3% the year before, even as the cost of acquiring that guest through the channel rose 25% between 2019 and 2025 against only a 19% rise in RevPAR. What did change was the terms: chains negotiated lower commissions and built loyalty programs that now anchor most of their direct volume. On the discovery side, meanwhile, something is moving much faster. U.S. traditional search-engine use for travel research fell from 51% to 36% in a single year while generative-AI tool use for travel more than doubled, from 6% to 15%, according to Phocuswright, a swing large enough that it's flagged as a single-study finding awaiting a second source rather than settled fact. Globally, OTA structure varies sharply by market too: India's OTAs handle 55% of the country's online gross bookings, and the top five OTA markets, the U.S., China, Japan, Germany, and India, together account for more than half of global OTA volume, so the pace and shape of this shift will not look identical everywhere.
What this does to how a traveler discovers, trusts, and chooses a place to stay is more complicated than 'AI is replacing search.' Pew Research Center's study of real Google search sessions found that when an AI summary appears, people click a traditional organic result in just 8% of visits versus 15% without one, click into the AI summary itself in only 1% of visits, and are more likely to abandon the session entirely, 26% versus 16%. Visibility inside an AI answer, in other words, is not the same as earning attention or a click. Trust still runs through more familiar mechanics too: a controlled experimental study found that positively framed reviews paired with an easy-to-read numerical rating measurably increase booking intention and trust, while leading with negative information early dampens both, a genuinely causal finding rather than a correlation. And AI adoption among travelers themselves skews toward busy, financially engaged travelers rather than the youngest cohort, with millennials adopting AI for travel planning at higher rates (58%) than Gen Z (45%) or boomers (11%), which should reset any assumption that this shift is a niche, generational one.
All of this redraws where a hospitality brand's attention actually sits, and by how much, in ways that current measurement tools weren't built to answer. OTAs remain the dominant marketplace surface for independent hotels and the primary revenue engine feeding a quarter of global travel spend; reviews remain a causal lever on trust and booking intent; search is still present but shrinking as a starting point; and the AI-answer layer is growing fastest of all while remaining the least measured. Independent properties are also the least prepared for that last shift, with only 41% reporting any AI adoption versus almost 80% of chains, which means the properties most reliant on the OTA marketplace are also the ones least equipped to be found in the layer growing fastest. No independent, hospitality-specific census of AI share-of-answer exists yet to tell a brand exactly where it stands in that layer today, which is precisely the terrain a Visibility Corpus is built to map: not one channel in isolation, but where a market's attention actually sits right now, and how fast it's moving.
The data, in one read
The channel that keeps taking a bigger cut
Global travel is a projected $1.67 trillion in gross bookings for 2025, and OTAs are projected to book $408 billion of that, roughly one in four travel dollars spent worldwide, per Phocuswright's Travel Forward report. That share is not distributed evenly across travel categories. Hotels do the heavy lifting inside the OTA channel: hotels account for 63% of U.S. OTA gross bookings, while only 13% of airline online revenue flows through an OTA at all, according to Phocuswright's U.S. Online Travel Agency Market Essentials research. There's a real inconsistency worth flagging here: Phocuswright's own reporting shows two figures for U.S. OTA gross bookings in the same year, $100.3 billion in the market-essentials product (about one-fifth of U.S. travel gross bookings, projected to reach 21% by 2028) and roughly $112 billion when the U.S. is ranked among the top five OTA markets in the Travel Forward report. Both are properly sourced and likely reflect different scope or timing, not an error, but neither should be quoted as if it settles the question on its own.
For independent properties, the reliance is deeper and getting deeper still. Cloudbeds' State of Independent Hotels study, analyzing 90 million bookings, found OTA share of independent-hotel bookings rose to 63.4% globally in 2025, up from 61.3% the year before, with North America climbing 3.3 points to 52.7%. The squeeze shows up on the cost side too: cost of guest acquisition rose 25% between 2019 and 2025, outpacing a 19% rise in RevPAR. An independent hotel is paying more to fill the same room through the same channel, and that channel's share of its bookings is growing, not shrinking.
Cost of guest acquisition rose 25 percent between 2019 and 2025, outpacing a 19 percent rise in RevPAR, while independent hotels routed a growing share of their bookings through the channel anyway.
A decade of trying to change that, and the scoreboard barely moved
Hilton's 2016 "Stop Clicking Around" campaign, which ran across 18 countries and reportedly cost around $100 million, was the clearest public shot fired in the direct-booking fight, according to Skift's retrospective. Ten years on, Skift reports that OTAs still control roughly the same overall share of hotel distribution industry-wide. That's not the same as saying nothing changed. What chains actually won over that decade was lower commissions, better contract terms, and loyalty programs that now anchor most of their direct volume, even where the headline OTA share never really moved.
IHG has stated that its loyalty members are about 10 times more likely to book direct and spend roughly 20% more on average than non-members, a figure worth reading as directional rather than as an industry constant. It comes from one company, self-reported through trade press, not from an independent audit, and shouldn't be generalized to every chain or program without its own corroboration.
The economics differ in another way worth noting: OTA-booked hotel stays cancel far more often than direct ones. About 22% of OTA bookings were canceled in Cloudbeds' 2025 study period, versus 10.6% of independent-hotel direct bookings, and average cancellation lead time lengthened to 39 days from 35 days in 2023. A direct booking is worth more in practice, not just on paper: it's more likely to actually happen.
Chains and independents are not fighting this on the same footing
AI adoption inside hotels is running at very different speeds depending on the size of the operation. An H2c report found only 41% of independent hotels reported using AI in some capacity, versus almost 80% of hotel chains. A separate Canary Technologies survey of 400 hospitality IT decision-makers found hotels expect AI's biggest near-term impact to land on guest communications (58%), search visibility (52%), personalization (51%), and direct bookings (48%); half of respondents said they were already using AI in some form, and 82% planned to increase usage over the next 12 months.
Put those two findings together and a pattern appears: the properties least equipped to adopt AI tools are the same ones already most dependent on OTAs, and the discovery layer growing fastest, generative AI, is exactly the one where adoption gaps compound fastest. That's not a forecast, it's what the current adoption numbers already show.
Meanwhile, the front door itself is moving
Independent of the OTA fight, where travelers start their research is shifting quickly. Phocuswright's Travel Forward report, citing its "Search Slips, AI Surges" research, found that U.S. traditional search-engine use for travel research fell from 51% in late 2024 to 36% in the second half of 2025, while generative-AI platform use rose from 6% to 15% over the same window; social network use for travel research rose more modestly, from 16% in 2023 to 19% in 2025. This is a single proprietary study reporting an unusually large one-year swing, without a second independent source confirming it yet, so it belongs in the emerging category: directional, not settled.
The adoption pattern itself cuts against the easy assumption that this is purely a youth trend. As of the second half of 2025, 58% of active U.S. travelers had used AI for something, and 39% specifically for travel planning, per Phocuswright. Adoption skewed toward busy travelers rather than the youngest cohort: 58% of millennials reported using AI for travel versus 45% of Gen Z and just 11% of boomers. A quarter to a third of travelers across the geographies Phocuswright surveyed said they're already interested in booking inside a generative-AI platform or letting an AI agent book on their behalf, though the actual transactional capability inside these products remains nascent.
Search engine use for travel research fell from 51 percent to 36 percent in a single year. Generative-AI use for travel more than doubled.
What happens when the AI answer actually shows up
The clearest independent data on what happens when a search engine hands back a synthesized answer instead of a list of links doesn't come from travel research at all, it comes from Pew Research Center's study of real Google search sessions. Pew tracked 900 U.S. adults across 68,879 searches, 12,593 of which produced an AI summary, and found that when an AI summary appears, people click a traditional organic result in just 8% of visits, versus 15% of visits when no summary appears. They click a link inside the AI summary itself in only 1% of visits, and they're more likely to end the search session without clicking anything at all: 26% versus 16%. About 18% of the searches Pew tracked produced an AI summary.
None of this data is hospitality-specific, but the implication for lodging brands is direct. Being present inside a synthesized answer is not the same thing as earning a click, and for a category where roughly a fifth of searches already surface a summary, that gap between visibility and traffic is exactly the kind of thing a lodging brand needs measured for its own market, not assumed from general search behavior.
When an AI summary appears in a Google search, people click a traditional link 8 percent of the time instead of 15 percent, and click into the AI summary itself in just 1 percent of visits.
Reviews still do real, causal work
The best-evidenced finding in this study is also the oldest. A controlled experimental study by Sparks and Browning, published in Tourism Management, used a full factorial design to test how the framing of online hotel reviews shapes booking intention and trust. Positively framed reviews paired with an easy-to-process numerical rating increased stated booking intentions and trust; leading with negative information early dampened both. Because it was a genuine experiment rather than a survey of stated preference, it's one of the few pieces of evidence in this study that supports a causal claim rather than just a correlation.
Some things couldn't be confirmed here too. Vendor content across the industry frequently cites a precise revenue or RevPAR lift tied to a marginal increase in review rating, often attributed to a 2009 study. That paper couldn't be independently re-verified in this research pass due to a paywall, so any specific lift figure circulating in sales material should be treated as unverified until the primary source is re-confirmed directly.
The category itself isn't going anywhere
It's easy to read a study like this and assume travel discovery is being upended wholesale. The demand for the category isn't in question: 44% of U.S. leisure travelers and 63% of business travelers choose hotels as their lodging, and nearly 80% of hotel-bound guests plan to stay at a midscale-or-higher property, according to an AHLA survey fielded by Morning Consult. The fight described in this study isn't about whether people want hotel rooms. It's about which door they walk through to find and book one, and that door is changing faster than most operators are tracking.
The measurement gap is the real opportunity
Here's the limit of everything above: nobody publishes an independent, hospitality-specific 'share of answer' for lodging brands inside ChatGPT, Gemini, or Google's AI Overviews, the way SEO rank-tracking has existed for classic search for two decades. Phocuswright, Pew, and the vendor surveys cited throughout this study measure general search and AI-tool adoption, or self-reported hotelier belief about AI's impact. None of them measure, independently and specifically, where a given hospitality brand actually stands inside the AI-answer layer today. That's the frontier, not a solved problem, and it's exactly where this study's evidence runs out.
A few other things stayed genuinely unsettled in this research and are worth naming: short-term and vacation-rental market sizing is contested, with comparable 2025-2026 estimates from different vendors ranging from about $104.6 billion to $165.7 billion to $149.2 billion depending on scope and methodology; regional behavior outside the U.S., U.K., France, and Germany on AI trip-planning adoption wasn't directly available in this research pass; and metasearch share-of-bookings claims (Google Hotel Ads, Trivago, TripAdvisor) came only from vendor and agency blogs rather than an authoritative primary source, so they were left out of this study's evidence entirely rather than reported at a lower tier.
The evidence, in numbers
Key findings, dated and sourced
-
Global travel market reached approximately $1.67 trillion in gross bookings in 2025 (projected), with growth moderating from the 2023-24 recovery surge.
established Phocuswright (Northstar Travel Group), Travel Forward: Data, Insights and Trends for 2026 (2026-07)
-
OTAs are projected to generate $408 billion in bookings in 2025, about one in four travel dollars spent worldwide; OTA power is highly asymmetric across lodging, air, and cruise/rail.
established Phocuswright, Travel Forward: Data, Insights and Trends for 2026 (2026-07)
-
U.S. OTA gross bookings rose 4% in 2025 to $100.3 billion, about one-fifth of U.S. travel gross bookings and projected to reach 21% by 2028; hotels account for 63% of U.S. OTA gross bookings while only 13% of airline online revenue flows through an OTA.
established Phocuswright, "13% Is the Number That Explains Why OTAs Need Hotels to Keep Working" (U.S. Online Travel Agency Market Essentials 2026) (2026-07)
-
OTA share of independent-hotel bookings rose to 63.4% globally in 2025, up from 61.3% in 2024, across 90 million bookings analyzed, with North America up 3.3 points to 52.7%; cost of guest acquisition rose 25% between 2019 and 2025, outpacing a 19% rise in RevPAR.
emerging Cloudbeds (reported by PhocusWire), State of Independent Hotels study, cited in "Independent hotel reliance on OTAs increased in 2025" (2026-03-25)
-
OTA-booked hotel stays cancel far more often than direct bookings: about 22% of OTA bookings were canceled in the 2025 study period versus 10.6% of independent-hotel direct bookings; cancellation lead time lengthened to 39 days from 35 days in 2023.
emerging Cloudbeds (reported by PhocusWire), State of Independent Hotels study, cited in "Independent hotel reliance on OTAs increased in 2025" (2026-03-25)
-
U.S. travelers' primary trip-research channels are shifting from search engines toward AI: use of traditional search engines for researching travel fell from 51% (late 2024) to 36% (2H 2025), while generative-AI platform use rose from 6% to 15% over the same window.
emerging Phocuswright, Travel Forward: Data, Insights and Trends for 2026, citing "Search Slips, AI Surges: Travel's New Front Door" (2026-07)
-
As of 2H 2025, 58% of active U.S. travelers used AI for something and 39% used it specifically for travel planning; adoption skews toward busy travelers rather than the youngest cohort (58% of millennials vs 45% of Gen Z vs 11% of boomers).
established Phocuswright, Travel Forward: Data, Insights and Trends for 2026 (2026-07)
-
A quarter to a third of travelers across surveyed geographies say they are already interested in booking travel inside a generative-AI platform or letting an AI agent book on their behalf, though transactional capability inside gen-AI products remains nascent.
emerging Phocuswright, Travel Forward: Data, Insights and Trends for 2026 (2026-07)
-
Google users who encounter an AI summary click a traditional organic result in only 8% of visits versus 15% without an AI summary; they click a link inside the AI summary itself in just 1% of visits and are more likely to abandon the search session (26% vs 16%); about 18% of searches produced an AI summary.
established Pew Research Center, "Google users are less likely to click on links when an AI summary appears in the results" (Athena Chapekis and Anna Lieb); n=900 U.S. adults, 68,879 searches, 12,593 with AI summaries, browsing collected April 2025 (2025-07-22)
-
A controlled experimental study found that how online hotel reviews are presented causally shapes both booking intention and trust: early negative information dampens both, while positively framed reviews paired with an easy-to-process numerical rating increase stated booking intentions and trust.
established Griffith University (Tourism Management journal, Elsevier), Sparks, B.A. & Browning, V., "The impact of online reviews on hotel booking intentions and perception of trust," Tourism Management, Vol. 32, Issue 6, pp. 1310-1323 (2011-12)
-
Hotels remain the dominant lodging choice for U.S. travelers: 44% of leisure travelers and 63% of business travelers select hotels, and nearly 80% of hotel-bound guests plan to stay at midscale-or-higher properties.
established American Hotel & Lodging Association (survey fielded by Morning Consult, n=2,202 U.S. adults, Sept 6-8, 2025), AHLA holiday travel survey press release (2025-09-30)
-
Hilton's 2016 "Stop Clicking Around" direct-booking campaign ran across 18 countries; ten years on, OTAs still control roughly the same overall share of hotel distribution, but chains have won lower commissions, better contract terms, and loyalty programs that now anchor most direct volume.
emerging Skift, "Direct Booking Tug-of-War: Hotels' Long Bid to Take Back Power" (2026-05-27)
-
IHG stated loyalty members are about 10 times more likely to book direct and spend roughly 20% more on average than non-members.
contested IHG (as reported by Skift), "Direct Booking Tug-of-War: Hotels' Long Bid to Take Back Power" (2026-05-27)
-
A survey of 400 hospitality IT decision-makers found hotels see AI's biggest near-term impact on guest communications (58%), search visibility (52%), personalization (51%) and direct bookings (48%); half reported already adopting AI, 82% planned to increase usage in the next 12 months.
emerging Canary Technologies (reported by PhocusWire), Canary Technologies hospitality AI survey, cited in "Independent hotel reliance on OTAs increased in 2025" (2026-03-25)
-
AI adoption inside hotels is far more advanced at large chains than independents: only 41% of independent hotels reported using AI in some capacity versus almost 80% of hotel chains.
emerging H2c (reported by PhocusWire), H2c hotel AI adoption report, cited in "Independent hotel reliance on OTAs increased in 2025" (2025-10)
-
India's OTAs account for 55% of the country's online gross bookings, outpacing supplier-direct channels; the top five OTA markets by 2025 projected gross bookings are the U.S. (~$112B), China ($60B), Japan ($24B), Germany ($18B) and India ($14B), together more than half of global OTA volume.
established Phocuswright, Travel Forward: Data, Insights and Trends for 2026 (2026-07)
Learning outcomes
What this study teaches
- Don't treat this as an 'OTA versus direct' fight alone. Independent-hotel reliance on OTAs is rising, and the cost of winning that guest is rising faster than the revenue it brings in.
- Loyalty infrastructure and better contract terms, not headline distribution share, are what chains have actually won over a decade of trying to reclaim direct bookings. That's the more realistic target for a smaller brand too.
- Guest research has already started migrating toward AI tools faster than any hospitality-specific measurement exists to track it, which means you can't yet get an objective read on where you stand there. That gap is worth closing before it widens.
- Independent properties are far behind chains on AI adoption, and that's the exact discovery layer growing fastest right now. Being behind on adoption where behavior is shifting fastest is a real risk, not a future one.
- Reviews still causally move booking intent and trust in controlled research, so how they're framed and presented matters as much as the star average itself.
Honest limits
What this does not yet settle
- No independent, hospitality-specific census of the AI-answer layer's share of voice exists yet. Phocuswright, Pew, and the vendor surveys cited here measure general search and AI-tool adoption, or self-reported hotelier belief about AI's impact, not a verified, third-party 'share of answer' for lodging brands inside ChatGPT, Gemini, or AI Overviews the way SEO rank-tracking exists for classic search. That is the frontier, not a solved measurement.
- Short-term and vacation-rental market sizing is genuinely contested: comparable 2025-2026 estimates from different research vendors ranged from about $104.6 billion to $165.7 billion to $149.2 billion depending on methodology and scope, so no single figure should be treated as settled without naming its source and scope.
- No rigorous, recent, peer-reviewed causal study was located quantifying the exact revenue or RevPAR lift from a marginal increase in review rating or review volume for hotels specifically. A frequently cited 2009 elasticity study could not be independently re-verified in this pass due to a paywall, so any such lift figure circulating in vendor material should be treated as unverified until the primary paper is re-confirmed.
- IHG's loyalty-to-direct-booking multiplier (10x) is a single company's self-reported figure relayed through trade press, not an independently audited or peer-reviewed statistic, and should not be generalized across the industry without corroboration from other chains or a neutral third party.
- Regional and non-English-language traveler behavior on AI trip-planning adoption, in markets like China, Southeast Asia, or the Middle East, was not directly retrieved in this research pass; the generational and geographic AI-adoption data available covers primarily the U.S., U.K., France, and Germany.
- Metasearch (Google Hotel Ads, Trivago, TripAdvisor) share-of-bookings claims found during this research came only from vendor and agency blogs rather than an authoritative primary source such as Google, STR, or a dedicated Phocuswright metasearch report, so they were left out of this study's evidence entirely rather than reported at a lower tier.
This is a synthesis of dated, attributed evidence, not a census. The AI-answer layer in particular has no independent, Nielsen-grade measurement yet, so readings of it are directional and named as a frontier, never presented as settled.
Straight answers
Frequently asked questions
Are OTAs losing ground to direct hotel bookings?
No, the opposite is happening for independent hotels. The Cloudbeds State of Independent Hotels study found OTA share of independent-hotel bookings rose to 63.4 percent globally in 2025, up from 61.3 percent in 2024, and the cost of acquiring that guest rose 25 percent between 2019 and 2025, outpacing a 19 percent rise in RevPAR. Even Hilton's high-profile 2016 "Stop Clicking Around" campaign didn't meaningfully move the industry-wide OTA share a decade later, per Skift; what chains actually won was lower commissions and loyalty programs that anchor direct volume, not a smaller OTA slice.
Is travel research really moving from Google to AI tools?
It looks that way, but the study flags it as an emerging finding, not a settled one. Phocuswright found U.S. traditional search-engine use for travel research fell from 51 percent to 36 percent in a single year while generative-AI tool use for travel more than doubled, from 6 percent to 15 percent, a swing large enough that the study itself notes it comes from one proprietary source awaiting a second one to confirm it.
If a hotel shows up in an AI answer, does that actually bring in guests?
Not on its own. Pew Research Center's study of real Google search sessions found that when an AI summary appears, people click a traditional organic result in just 8 percent of visits versus 15 percent without one, click into the AI summary itself in only 1 percent of visits, and are more likely to abandon the search entirely, 26 percent versus 16 percent. Being present inside an AI answer is not the same as earning a click, which is exactly why visibility there needs to be measured separately from traffic.
Do online reviews actually change whether someone books a hotel?
Yes, and this is the most solidly proven finding in the study. A controlled experimental study by Sparks and Browning, published in Tourism Management, found that positively framed reviews paired with an easy-to-read numerical rating measurably increased booking intention and trust, while leading with negative information early dampened both. Because it was a real experiment rather than a survey of stated preference, it supports a causal claim, not just a correlation.
Are independent hotels ready to compete in AI-driven search?
The evidence says no. An H2c report found only 41 percent of independent hotels reported using AI in any capacity, versus almost 80 percent of hotel chains, and no independent, hospitality-specific measurement of AI share-of-answer exists yet for anyone to check where they stand in that layer. The properties most reliant on OTAs are also the ones least equipped to be found in the discovery channel the study describes as growing fastest, which the study calls a real, current gap rather than a future risk.
Provenance
References
- Phocuswright (Northstar Travel Group), "Travel Forward: Data, Insights and Trends for 2026" https://www.phocuswright.com/Travel-Research/Research-Updates/2026/Travel-Forward-Data-Insights-and-Trends-for-2026
- Phocuswright, "13% Is the Number That Explains Why OTAs Need Hotels to Keep Working" (U.S. Online Travel Agency Market Essentials 2026) https://www.phocuswright.com/Travel-Research/Research-Updates/2026/the-number-that-explains-why-otas-need-hotels-to-keep-working
- PhocusWire (Linda Fox), "Independent hotel reliance on OTAs increased in 2025," citing Cloudbeds' State of Independent Hotels study, the Canary Technologies hospitality AI survey, and the H2c hotel AI adoption report https://www.phocuswire.com/news/distribution/hotel-distribution-ai-trends-cloudbeds-canary
- Pew Research Center (Athena Chapekis and Anna Lieb), "Google users are less likely to click on links when an AI summary appears in the results," July 22, 2025 https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- Sparks, B.A. and Browning, V., "The impact of online reviews on hotel booking intentions and perception of trust," Tourism Management, Vol. 32, Issue 6 (2011), pp. 1310-1323 https://www.sciencedirect.com/science/article/abs/pii/S0261517711000033
- American Hotel & Lodging Association, holiday travel survey (fielded by Morning Consult, n=2,202), September 2025 https://www.ahla.com/news/half-americans-plan-travel-holiday-season-hotels-remaining-top-choice-lodging
- Skift, "Direct Booking Tug-of-War: Hotels' Long Bid to Take Back Power," May 27, 2026 https://skift.com/2026/05/27/direct-booking-tug-of-war-hotels-long-bid-to-take-back-power/
Every measured figure is dated to its capture and tagged with an evidence tier. Every cited work is real and locatable. Where an engine could not be captured this round, it is named as uncaptured, not estimated. Small-sample readings are labelled as directional.