# The Future of Short-Term Rental Booking Beyond the OTA > Airbnb has begun testing something it had never previously priced: the difference between a guest Airbnb finds and a guest the host brings, with selected hosts reporting shareable booking links at a service fee of roughly 6% to 10% rather than the 15.5% most hosts now pay. The reservation, payment, and AirCover stay on Airbnb; only the origin of the traveler changes. Open-web booking infrastructure is improving at the same moment Airbnb's share of professionally managed U.S. reservations is rising. - Published: 2026-08-30 - Last reviewed: 2026-08-30 - Author: Kate Swanson, Founder, Book With Haven - Canonical: https://www.bookwithhaven.com/research/the-future-of-short-term-rental-booking-beyond-the-ota - Cite as: Kate Swanson. (2026). The Future of Short-Term Rental Booking Beyond the OTA. Haven Research. https://www.bookwithhaven.com/research/the-future-of-short-term-rental-booking-beyond-the-ota ## Key findings - As of August 29, 2026, Skift reported that Airbnb is testing host-shared booking links with host fees of roughly 6% to 10% against a standard 15.5%, while checkout remains on Airbnb. Airbnb has published nothing about the program. - Key Data's professionally managed U.S. panel showed Airbnb at 50% of Q1 2026 reservations, up from 46% a year earlier, while direct fell to 23% from 26%. - Google Vacation Rentals can already route travelers from Search, Maps, or Google Travel to a partner booking page and states there are no fees for those referrals, but access is gated behind Hotel Center and an Early Adopters Program. - Phocuswright reported 56% of active U.S. travelers had used AI for at least one trip in the prior year as of early 2026; Google launched hotel booking inside AI Mode on August 27, 2026. - Phocuswright puts the 2025 U.S. split of online hotel bookings at roughly 52% OTA and 48% supplier-direct, and expects that split to hold through 2029. ## Report **Research date: August 30, 2026.** **Fact-check date: August 30, 2026.** Primary geography is the United States, with European and global evidence where it clarifies the picture. This report is a research foundation for Book With Haven. Findings are classified as established fact, supported interpretation, early signal, or prediction. Every numbered citation was checked against the live source on August 30, 2026. ## Executive summary The infrastructure for open-web booking is improving at the same moment Airbnb's share of actual short-term rental reservations is rising. That is a stronger and more honest premise than "Airbnb is being disrupted," and it sets up a genuine question rather than a foregone conclusion.
Finding Classification What the evidence establishes
Airbnb is testing reduced service fees when hosts source the guest themselves. Established fact, single source Selected hosts received shareable "direct booking links" for distribution through social media, email, and message boards. Reported host fees run 6% to 10% against a standard 15.5%. Checkout, payment, and the reservation remain on Airbnb.
The program is host-sourced acquisition, not direct booking. Supported interpretation Airbnb retains the transaction and the guest relationship. What changes is who originated the demand.
That distinction is the most economically interesting part of the experiment. Supported interpretation By charging less for a guest the host supplies, Airbnb implicitly separates the value of demand generation from the value of payments, checkout, support, trust, and protection.
Nothing indicates a broad rollout. Early signal The program is a limited pilot. Eligibility, geography, the reason some hosts see 6% and others 10%, and any expansion plan are all undisclosed. Airbnb has issued no public statement.
Direct booking is not free booking. Established fact An independent reservation still carries payment processing, software, fraud exposure, service, cancellation, and marketing costs. Stripe's published U.S. standard rate for domestic online cards is 2.9% plus $0.30, before any acquisition spend.
The strongest evidence against a simple direct-booking narrative is current channel share. Established for one dataset Key Data's professionally managed U.S. panel showed Airbnb at 50% of reservations and 37% of revenue for Q1 2026, against 23% of reservations and 35% of revenue for direct. Direct fell from 26% of reservations a year earlier; Airbnb rose from 46%.
Google can already send vacation-rental shoppers to a host's own booking page at no referral cost. Established fact Vacation rental listings appear across Search, Maps, and Google Travel, and free booking links redirect travelers to the partner's own site. Google states there are no fees for Google-generated referrals or bookings.
AI has become a real travel-discovery channel, though measurement varies sharply by definition. Established trend Phocuswright reported 56% of active U.S. travelers had used AI for at least one trip in the prior year as of early 2026, and separately put standalone generative AI platforms at 33% for trip research. Deloitte's narrower 2025 trip-planning measure was 15%.
Hotels offer the best available precedent, and the precedent is coexistence. Supported interpretation Phocuswright puts the 2025 U.S. split of online hotel bookings at roughly 52% OTA and 48% supplier-direct, and expects it to hold through 2029.
Two limits deserve to be stated plainly. First, the Airbnb pilot rests on a single published source. Skift reported it on August 29, 2026, working from messages hosts circulated in Facebook groups and on LinkedIn. Airbnb has published nothing about the program, no other outlet has covered it independently, and an earlier draft of this report cited a corroborating Airbnb Community forum post that could not be located. Every claim about the pilot should be read as "as reported by Skift." Second, several load-bearing market statistics come from paywalled or subscription-distributed research and are available publicly only through secondary reporting. Those are flagged in the source library. The conclusion the research will support is that the next credible direct-booking company will not win by offering hosts a cheaper place to take a reservation. Checkout is the commodity layer. It will win only if it solves the problem Airbnb still solves better than anyone: putting the right property in front of the right traveler, generating enough trust to convert, and doing so repeatedly at an acquisition cost below the commission being displaced. ## 1. Airbnb's host-sourced link experiment ### What is actually being tested Skift reported on August 29, 2026 that Airbnb had begun issuing "direct booking links" to selected hosts. The message Airbnb sent participants was titled "Share your link, get a lower fee," and instructed hosts to distribute the links through their own channels, naming social media, email, and message boards. A traveler who follows the link still books on Airbnb. The reservation retains ordinary platform benefits, AirCover included. Skift reported participating-host fees of 6% to 10% against a standard 15.5%. [1] Haven Insights covered the same pilot for hosts in [Airbnb Direct Booking Links: What Hosts Need to Know](/blog/airbnb-direct-booking-links). This report treats the experiment as evidence about how Airbnb prices demand, not as product news. Two qualifications matter more than they might appear to. Skift attributes the 6% to 10% range as reported, not as stated by Airbnb, and its sourcing consists of messages hosts shared in Facebook host groups and on LinkedIn. The article carries no Airbnb comment or confirmation. Nothing published by Airbnb defines this program. Its newsroom has run nothing on it, no Help Center article describes it, and as of August 30 no other outlet, including PhocusWire, Reuters, Bloomberg, and the major technology press, has covered it independently. [1][5] An earlier draft of this research cited an Airbnb Community forum post reproducing a participant invitation email with a 6% fee and a pilot disclaimer. That post could not be located on re-verification and has been removed. The 6% figure now rests entirely on Skift's secondhand sourcing. Any published claim about this pilot should be attributed to Skift explicitly. ### What remains unknown
Question What is known Confidence
Who receives access Selected hosts have been invited. High
Geography Not established by any public source. Low
Fee Skift reports a 6% to 10% range. High that a range exists, low on segmentation
Why some hosts see 6% and others 10% Undisclosed. Unknown
Where the booking occurs On Airbnb. Very high
Who sources the guest The host, through a link the host distributes. Very high
AirCover Skift reports normal benefits including AirCover are retained. High
Whether Airbnb markets these bookings for the host The documented flow begins with a link the host distributes. No separate acquisition component has been identified. High for the documented pilot; do not generalize
Whether Airbnb will expand it No reliable evidence either way. Unknown
The flow is easier to reason about drawn than described:
Diagram of Airbnb's shareable booking link pilot. On the left, three outlined steps: the host receives an Airbnb booking link, distributes it through social media, email and message boards, and the guest clicks it. A handoff divider separates those from three solid Airbnb steps on the right: Airbnb checkout and payment, Airbnb reservation, and AirCover, support, reviews and the guest relationship. The reported host fee is 6% to 10% against a standard 15.5%. What changed is who found the guest. What did not change is where the booking lives.
As reported by Skift, August 2026. Airbnb has published nothing about this pilot.
### Why it matters more than its size suggests Airbnb is experimenting with acquisition-sensitive pricing. When Airbnb supplies the demand, most hosts on its single-fee structure pay 15.5%. When a participating host supplies the prospect, Airbnb is reportedly willing to accept substantially less while still providing the entire transaction stack. [1][2] That supports a defensible line: > Airbnb's own pricing now distinguishes between a booking it originates and a booking the host originates. It would not be defensible to write that Airbnb has conceded its "real cost for payments and protection is 6%." Experimental pricing routinely includes strategic discounts, and a promotional rate reveals nothing reliable about underlying cost structure. There is also a policy tension worth noting without overstating. Airbnb's Off-Platform and Fee Transparency Policy prohibits hosts from including links that take people off the Airbnb platform in listings or messages, from asking or encouraging users to move current, future, or repeat bookings off Airbnb, from offering or soliciting discounts to book off-platform, and from selling, sharing, or using guest contact information for marketing communications or contact lists. [3] The pilot does not violate that policy, because it routes travelers back to Airbnb rather than away from it. No published Airbnb text reconciles the two, and this report does not imply a carve-out that Airbnb has never announced. ## 2. OTA economics ### What hosts currently pay Airbnb's Help Center states that under the single-fee structure most hosts pay 15.5%, remaining hosts typically pay 14% to 16%, and listings in Brazil and Mexico pay 16%. Under the older split-fee structure, most hosts pay 3% while guests pay 14.1% to 16.5% of the booking subtotal, with a 4% host fee in Brazil and Mexico. The single fee is mandatory for traditional hospitality listings such as hotels and serviced apartments, for hosts using property management software, and for hosts in countries subject to that structure. [2] Connecting a listing to a PMS moves a host onto the higher host-paid fee. The current published schedule is also summarized in [Airbnb Host Fees Explained](/blog/airbnb-host-fees-explained). Vrbo's published pay-per-booking terms are a 5% commission plus a 3% payment processing fee. The commission applies to the rental amount plus additional charges such as cleaning and pet fees; the processing fee applies to total payments including taxes and refundable damage deposits. Australia, Japan, and New Zealand are exceptions, charging the 5% commission plus GST with no separate processing fee. [9] Booking.com publishes no universal commission rate. Its partner documentation states that the percentage varies by country and can vary by property type or location, and that a higher-than-contracted percentage generally indicates participation in a program such as Genius, Preferred Partner, or Visibility Booster. [10] Preferred Partner offers greater search visibility in exchange for what Booking describes as a small increase in commission, with the exact figure unpublished. [11] Any article quoting a single global Booking.com rate is guessing. Regulatory scrutiny of exactly that mechanism is live. On April 22, 2026, Italy's AGCM opened an investigation into Booking.com entities over alleged unfair commercial practices, focused on the Preferred Partners programme and the regulator's contention that selection favors partners offering higher commissions. [13] It is a consumer-protection action rather than a pure competition case, and an investigation is not a finding.
Channel or model Published host-side cost What it covers Qualification
Airbnb single fee 15.5% for most hosts Marketplace demand, checkout, payment, support, trust, reviews, protections 14% to 16% for remaining hosts; 16% in Brazil and Mexico [2]
Airbnb split fee 3% host, 14.1% to 16.5% guest Same platform Being retired for several host categories [2]
Airbnb host-sourced link pilot Reported 6% to 10% Airbnb checkout and normal benefits while the host supplies traffic Limited pilot, single source, segmentation unknown [1]
Vrbo pay-per-booking 5% commission plus 3% processing Marketplace and transaction stack AU, JP, NZ differ; software-connected terms vary [9]
Booking.com Contractual, undisclosed Distribution and transaction services No published global rate; visibility programs raise it [10][11]
Independent checkout via Stripe 2.9% plus $0.30, U.S. domestic online card Payment processing only Excludes acquisition, software, support, fraud losses, insurance [15]
### The demand Airbnb is actually selling Airbnb states that its service fee helps cover services including 24/7 customer support. [2] The platform also supplies payment infrastructure and AirCover for Hosts, which includes guest identity verification, reservation screening that analyzes hundreds of factors and blocks bookings showing high risk of parties or damage, up to $3 million in host damage protection, $1 million in host liability insurance, and a 24-hour safety line, all subject to extensive terms and exclusions. [4] Scale is the harder asset to reproduce. Airbnb reported more than 5.5 million hosts and more than 2.5 billion cumulative guest arrivals in its Q2 2026 shareholder letter, alongside 148.3 million Nights and Seats Booked and $27.2 billion of gross booking value for the quarter. [7][8] One caution for anyone quoting that figure: Nights and Seats Booked now combines nights for stays with seats for services and experiences, so it is not comparable to the older lodging-only metric. Airbnb spent $2.59 billion on sales and marketing in 2025 against $12.24 billion of revenue. [6] Consumer demand at that scale is neither free nor casually reproducible. That evidence supports a sharper version of the direct-booking thesis than the usual formulation: > The booking form is replicable. Demand, trust, conversion, risk management, and repeat usage are the expensive parts. ## 3. Direct-booking economics A $1,000 reservation makes the arithmetic legible.
Cost Amount
Airbnb at 15.5% $155.00
Airbnb pilot at 10% $100.00
Airbnb pilot at 6% $60.00
Stripe U.S. domestic online card, 2.9% plus $0.30 $29.30
Horizontal bar chart of what a host pays on a $1,000 reservation. Airbnb's standard 15.5% fee costs $155.00. Airbnb's reported pilot range costs $100.00 at 10% and $60.00 at 6%. Booking direct on Stripe at 2.9% plus $0.30 costs $29.30. A bracket marks a $125.70 differential between the Airbnb standard fee and the Stripe cost. The chart shows fee differential only and excludes customer acquisition cost.
Fee differential only. Excludes customer acquisition cost, software, fraud, service and cancellation costs. Airbnb pilot rates as reported by Skift, August 2026, and not confirmed by Airbnb.
Against Airbnb's standard 15.5%, an independent booking saves roughly $125.70 on a $1,000 transaction before customer acquisition, software, fraud, service, insurance, chargebacks, refunds, and operational overhead. Against the reported pilot rates the gap narrows to about $70.70 and $30.70. These are arithmetic illustrations of fee differentials, not profit estimates. [1][2][15] The honest framing: **Net economics of direct acquisition = avoided OTA commission − payment cost − incremental customer acquisition cost − software cost − incremental servicing and risk cost.** Two of those terms are knowable in advance. Three are not, and the third one is where most direct-booking marketing quietly stops counting. There is a further constraint that rarely appears in vendor content. Airbnb's Off-Platform Policy bars using guest contact information for marketing communications or contact lists. [3] A host cannot convert Airbnb's acquired audience into an owned marketing database. Owned demand has to be acquired independently, or collected with appropriate consent through a channel the host controls. That single restriction is why the CRM and first-party data layer matters more than it looks like it should. ## 4. How travelers actually discover accommodation Expedia Group's Path to Purchase research remains the most useful public picture of the pre-booking journey, provided its methodology is described accurately. Published July 25, 2023, it combines two separate studies rather than one panel. A digital behavioral panel of more than 70,000 participants, tracked across desktop and mobile, produced the finding that travelers viewed an average of 141 travel content pages in the 45 days before booking, with U.S. travelers viewing up to 277 pages and more than five hours of travel content. A separate survey conducted by Luth Research between March 24 and April 19, 2023, with 5,713 adults across seven markets, produced the channel figures: OTAs appeared in 80% of research paths, search engines in 61%, social media in 58%, airline sites in 54%, and metasearch in 51%. [16] The study is Expedia-sponsored and its data is now three years old. It should be cited as large-scale industry research rather than neutral academic evidence. What it establishes is not a precise percentage but a structural fact: accommodation discovery is not a single-site behavior. Expedia's 2025 Traveler Value Index, published May 20, 2025 and surveying more than 11,000 consumers across 11 markets, found 61% of travelers finding trip ideas on social platforms, up from 35% in 2022, and roughly three-quarters willing to pay more for lodging with better reviews. [17] Phocuswright's narrower measure tells a usefully different story. Its U.S. figure for social network use in travel research rose from 16% in 2023 to 19% in 2025. [18] The gap between 61% and 19% is not a contradiction. "Used social media for travel inspiration" and "used social media to research or select a travel component" are different questions, and conflating them is the fastest way to lose credibility with a sophisticated reader. The claim the evidence actually supports is that travelers move through multiple discovery environments before transacting, and that the environment where a trip is imagined is frequently not the environment where it is booked. [16][18] ## 5. Google and open-web travel discovery ### The infrastructure already exists Google's vacation rentals product is the clearest existing evidence for the open-web premise. Google states that vacation rental listings appear alongside organic results, that travelers discover them through keyword searches, destination browsing in Maps, and lodging filters in Google Travel, and that partners sending accurate rates, availability, photos, descriptions, amenities, and reviews can build information-rich listings that link to their own landing pages. On cost, Google's language is direct: free booking links redirect users to book directly on the partner's website, and there are no fees for Google-generated referrals or bookings. [26] That architecture is close to the one this thesis anticipates. The practical path for a small host is still gated; [How to List on Google Vacation Rentals as a Small Host](/blog/google-vacation-rentals-small-hosts) walks through eligibility and connectivity partners. The transaction question is then: after discovery on Google, AI, social, paid, or organic surfaces, does checkout occur on an OTA, on a host-controlled site, or inside an AI or agent flow where the hotel or booking platform remains merchant of record?
Three-band diagram. The top band, labeled discovery and fragmenting, holds eight equally weighted channels: Google Search, Google Maps, AI assistants, social platforms, paid search, email and CRM, repeat guests, and OTAs. Fine lines run from all eight through a middle band reading the traveler decides where to book. The bottom band, labeled transaction and still concentrated, holds three heavier destinations: OTA checkout, where the platform owns the guest; host-controlled checkout, where the host owns the guest; and AI or agent checkout, marked emerging and hotels only as of August 2026, where the supplier or platform remains merchant of record.
OTAs appear in the discovery band as one channel among many. The transaction band has not fragmented the way discovery has.
### The gate on that infrastructure Access is the constraint, and it is a real one. Google's `VacationRental` structured data documentation, last updated December 10, 2025, states that the instructions are intended for sites that have already connected with a Google technical account manager and have access to Hotel Center, that the feature is limited to sites meeting eligibility criteria with additional integration steps required, and that completing the interest form is an expression of interest that does not guarantee an invitation into the Early Adopters Program. [28] Google separately maintains a published list of more than sixty approved vacation rental connectivity partners by region. [27] Adding JSON-LD to a property page does not produce a Google Vacation Rentals listing. The strategic opportunity for a platform is therefore not schema generation. It is becoming the connectivity layer that can satisfy feed, availability, pricing, landing page, policy, identity, and Hotel Center requirements across an entire portfolio of independent properties. ### The hotel precedent Phocuswright's U.S. Hotel & Lodging Travel Market Essentials 2026, published June 2026, puts the split within online hotel bookings at 52% OTA and 48% hotel websites and apps for 2025, and expects it to hold steady through 2029. [43] The scope qualifier matters: that is a share of online bookings, not of all hotel bookings.
A single stacked bar showing the split of US online hotel gross bookings in 2025: 52% online travel agencies and 48% hotel websites and apps. A dashed extension to 2029 is annotated that Phocuswright expects the split to hold. The headline reads that two decades into online hotel distribution, neither channel won.
Share of US online hotel gross bookings, 2025. Phocuswright, U.S. Hotel and Lodging Travel Market Essentials 2026. Online bookings only, not total hotel bookings.
The United Kingdom shows a different and, for this thesis, more encouraging pattern. Phocuswright reports supplier-direct at 78% of the U.K. online travel market in 2025, rising to 81% by 2029, with OTA share easing from 20% to 19% while OTA gross bookings still hit a record £10.3 billion in 2025. [44] Share can decline while absolute volume sets records. What hotels establish is coexistence rather than displacement. A mature digital travel category can sustain large aggregators for comparison and incremental demand, strong supplier-direct channels for brand searches and repeat guests, and search and metasearch layers sitting above both. Booking Holdings reinforces the point from the other direction. Its FY2025 Form 10-K reports $8.19 billion of marketing expense against $26.92 billion of revenue, roughly 30%, and states that a significant portion of consumer traffic derives from third-party platforms including Google and other search engines, mobile operating systems, app marketplaces, and mapping services. [12] > Even one of the world's largest travel marketplaces has to keep buying discovery from platforms sitting upstream of its own marketplace. That observation supports the case for an independent discovery layer without pretending OTAs are becoming irrelevant. ## 6. AI and travel discovery A companion Haven Research report, [AI, Distribution, and the Future of Direct Versus Intermediated Lodging Booking](/research/ai-distribution-and-the-future-of-direct-versus-intermediated-lodging-booking), examines whether AI changes who captures the lodging transaction. This section stays with adoption, click-through, and the first transactional integrations. ### Adoption The pace of change is the story. Phocuswright reported in March 2026 that 56% of active U.S. travelers had used AI to plan, book, or navigate at least one trip in the previous twelve months, against 43% in the second half of 2025 and 33% in the first half. [19][20] By July 2026 it reported standalone generative AI platforms such as ChatGPT at 33% usage for trip research, roughly five times their 2024 level. [21] Two cautions. Phocuswright does not publicly disclose the sample size behind the 56% figure, so no article should assert one. And a separate Phocuswright research update reports 58% of active U.S. travelers using AI for at least one purpose and 39% using it for travel research and planning, which is a different cut from the same broad program. [18] Cite the wave, not just the number. Deloitte's series is more conservative and directionally consistent. Its 2025 summer travel survey, published May 20, 2025 and fielded in late March and early April with samples of 1,794 and 1,064, found 15% of travelers using generative AI in trip planning, up from 10% the prior year. [24] Its holiday research, published November 12, 2025 and fielded September 26 to October 3 with 3,896 respondents of whom 2,099 qualified as holiday travelers, found 24% expecting to use generative AI, with 54% of those users expecting to research accommodations, placing accommodations third behind activities at 67% and destinations at 56%. [25] That 54% represents roughly 13% of holiday travelers overall, and the distinction is worth preserving. The variation between Phocuswright and Deloitte reflects definitions, samples, and fielding periods. Direction is the reliable signal. A stitched-together growth curve across incompatible methodologies is not. ### AI has not eliminated the click Phocuswright's March 2026 research found that 51% of travelers who used AI within search engines subsequently clicked through to source websites, which its analysts framed as a challenge to the zero-click narrative. [23] The scope is specific: travelers using AI inside search engines, not all AI users. The figure appears in secondary reporting rather than on Phocuswright's own public pages, and should be attributed accordingly. ### AI is already transacting On August 27, 2026, Google announced hotel booking inside AI Mode, rolling out in the U.S. in English. Travelers can review details including cancellation policy and complete the booking through Google Pay, while the hotel or booking platform acts as merchant of record and handles customer service. The named launch partners are Booking.com, Choice Hotels International, Expedia, Hilton, Hotels.com, IHG Hotels & Resorts, Marriott International, Priceline, Trip.com, and Wyndham Hotels & Resorts. [33] Google had laid the groundwork the previous November with an AI Mode travel canvas that assembles real-time flight and hotel data, Maps photos and reviews, and information from sites across the web into a single planning surface. [34] This cuts both ways. The optimistic reading is that AI creates a discovery layer capable of reasoning over inventory and matching highly specific traveler intent to an individual property, which is a task conventional keyword search performs badly. [34] The pessimistic reading is that the first transactional integrations went to the largest chains and OTAs, not to independent suppliers. If independent short-term rental inventory does not become equally standardized, real-time, and trustworthy in machine-readable form, AI could deepen incumbent distribution power rather than dilute it. Phocuswright's finding that recognized brands rank third at 32% among the factors driving travelers to act on AI recommendations, behind price comparisons at 44% and summarized reviews at 33%, points the same direction. [21][33] ## 7. The ecommerce analogy Google opened Shopping to free product listings on April 21, 2020 and extended free retail listings into the main Search results page on June 29, 2020. [35][36] In July of that year it made Buy on Google commission-free. [37] The architecture Google described was explicit: consumers discover products through Google and complete the purchase either on Google or on the retailer's own site. The machinery has grown considerably. Google now recommends both product structured data and Merchant Center feeds so retailers can expose price, availability, and product identity across Google surfaces, and states that providing both maximizes eligibility. [39][40] The Shopping Graph, which Google describes as the world's most comprehensive catalog, holds more than 60 billion product listings as of May 2026 and underpins AI Mode's shopping experience. [38] The strong half of the analogy: > Standardized, machine-readable inventory lets a discovery layer aggregate products from independent sellers without any single marketplace owning every transaction. The weak half is the claim that Google beat marketplaces. It did not. Google's commerce infrastructure grew alongside Amazon and other marketplaces rather than displacing them. Accommodation is also structurally harder than retail. Inventory expires nightly. Price varies by dates, length of stay, and party composition. Cancellation terms are material to the purchase decision. Regulatory eligibility varies by municipality. Fraud exposure is higher, and review history carries unusual weight. Google's own decision to gate vacation rentals behind Hotel Center connectivity, when product listings require no equivalent relationship, is itself evidence of that complexity. [26][28] A defensible formulation: Google made ecommerce discovery more distributed, and travel may distribute the same way without becoming marketplace-free. ## 8. Occupancy and channel distribution ### There is no single Airbnb occupancy rate Airbnb publishes Nights and Seats Booked, gross booking value, and related marketplace metrics, but not the denominator of bookable nights required to compute a host occupancy rate. [8] Third-party estimates therefore rest on different definitions, samples, and methods. AirDNA calculates property-level occupancy as reserved days divided by total active listing nights over a trailing twelve months, and market-level occupancy as nights booked divided by nights available to be booked within a month. A night counts as an active listing night only if it is unblocked, either reserved or available, and the listing has had a reservation within the past 28 days. [48] Listings that block an entire calendar while remaining live are excluded. That definition alone makes casual comparison against a host's own calendar, or against hotel occupancy, unreliable. AirDNA's own published figures illustrate the volatility. It gives a U.S. average Airbnb occupancy rate of 54.3%, with a seasonal range running from 41% in January to 67.5% in July. [49] A twenty-six point seasonal swing is a strong argument against quoting any single national number as meaningful for a specific property in a specific market.
Range chart of US Airbnb occupancy. A band runs from 41% in January to 67.5% in July on a scale truncated to begin at 30%, with an orange diamond marking the US average of 54.3%. The headline reads that there is no single Airbnb occupancy rate, and 26.5 points separate January from July before accounting for market, property type or availability.
AirDNA, US Airbnb occupancy, page last updated May 2026. Airbnb publishes no bookable-night denominator, so no official occupancy metric exists.
### Channel mix is the more useful measure Key Data's quarterly index for professionally managed U.S. properties reported Airbnb at 50% of reservations and 37% of revenue in the first quarter of 2026, direct at 23% of reservations and 35% of revenue, and Vrbo at 20% and 24%. Year over year, Airbnb rose from 46% of reservations and 34% of revenue while direct fell from 26% and 39%. [45]
Channel Reservation share, Q1 2026 Revenue share, Q1 2026
Airbnb 50% 37%
Direct 23% 35%
Vrbo 20% 24%
Other 7% —
Two grouped horizontal bar charts comparing share of reservations with share of revenue for US professionally managed short-term rentals in the first quarter of 2026. Share of reservations: Airbnb 50%, up 4 points year over year; direct 23%, down 3 points; Vrbo 20%, unchanged; other 7%, down 1 point. Share of revenue: Airbnb 37%, up 3 points; direct 35%, down 4 points; Vrbo 24%, unchanged; other 4%, up 1 point. A call-out notes that direct takes 35% of revenue on 23% of reservations.
Key Data index, Q1 2026 US data. Panel of professionally managed, PMS-connected operators, not a market census. Reported via The Host Report, April 2026.

U.S. reservation share, Key Data professionally managed panel, Q1 2026. "Other" is the residual after Airbnb, direct, and Vrbo. Revenue share for "Other" was not published in the public account.

Three qualifications belong wherever these figures are used. The report is labeled Q2 2026 and reports finalized Q1 2026 data. Key Data's index labeled Q1 2026 reports Q4 2025 channel mix, with materially different numbers: Airbnb 54% of reservations and 45% of revenue, direct 21% and 28%. [46] Citing the wrong edition produces the wrong figures. The public account of the Q1 2026 data comes from a single trade outlet. The primary PDF is distributed by email subscription and was not publicly retrievable, and no other outlet covered it. [45] The panel is self-selected. Key Data draws verified first-party reservation data directly from property management systems across a network it describes as 17,000-plus property managers, combined with OTA intelligence. [47] That is a large sample of professionally managed, PMS-connected inventory, not a census of the U.S. short-term rental market, and individual hosts are systematically underrepresented. With all three qualifications applied, the finding still stands and still cuts against the thesis. Among professional operators with the tooling and sophistication to run a direct channel, direct booking lost share to Airbnb over the past year. That is the problem a direct-booking platform would have to solve, stated in the industry's own data. ## 9. Arguments supporting the thesis The strongest supporting evidence does not show direct booking winning. It shows the preconditions for more distributed booking arriving. Airbnb's pilot is itself evidence that acquisition source carries distinct economic value. A host-sourced lead is being priced below a marketplace-sourced booking while retaining the same transaction stack. [1] Google has built a functioning, no-referral-fee route from accommodation discovery to a supplier-controlled booking page, and says so in its own documentation. [26] AI has become a travel research interface at speed, and roughly half of travelers using AI inside search engines still continue to source websites rather than stopping at the generated answer. [19][23] Hotels prove that supplier-direct can hold close to half of online distribution in a category with formidable OTAs, and the U.K. data suggests supplier-direct share can grow rather than merely persist. [43][44] Direct reservations in Key Data's panel contribute disproportionate revenue relative to their reservation count, which indicates the channel carries real economic value when an operator can actually acquire it. [45] And the marketplaces themselves demonstrate that demand is rented, not owned. Booking Holdings spent $8.19 billion on marketing in 2025 and Airbnb $2.59 billion, much of it flowing to platforms neither company controls. [6][12] ## 10. The strongest arguments against The most serious counterargument is empirical rather than theoretical. Airbnb is gaining reservation share in the best available dataset, rising four percentage points year over year while direct fell three. [45] Scale compounds. More inventory attracts guests, guest traffic attracts hosts, bookings generate reviews, reviews build trust and lift conversion. Airbnb also bundles payments, support, identity verification, reservation screening, and protection that an independent operator must either replace or knowingly forgo. [4][7] OTAs solve a problem individual property sites structurally cannot. A traveler searching one destination can compare dozens of properties, dates, prices, reviews, and cancellation terms in a single session. A host site starts every session with an audience of one and nothing to compare against. Paid acquisition can erase the commission advantage entirely. A single-property host bidding on a competitive destination query has none of Booking's conversion data, brand recognition, cross-property inventory, attribution infrastructure, or lifetime-value modeling. [12] Google may function as an intermediary rather than an open pipe. Its first transactional AI Mode integrations went to the largest chains and OTAs, not to independent suppliers, and its vacation rentals program is gated behind Hotel Center connectivity. [28][33] AI does not inherently democratize recommendation. Brand recognition ranks among the top three factors driving travelers to act on AI suggestions. [21] Which produces the most defensible version of the thesis: > Hosts will not reduce OTA dependence by owning a booking page. They reduce it only if independent distribution becomes competitive at acquisition, matching, trust, conversion, and retention. That is a harder claim to make and a much better one to defend. ## 11. Implications for Book With Haven Pursuing OTA-scale demand without a percentage commission requires substantially more than a direct-booking website. The capability list below separates what the research establishes from what remains strategic ambition.
Capability Why it matters Evidence
Canonical property identity Every property needs a persistent, indexable URL and entity data that Google and AI systems can interpret. Google structured data guidance [29][30]
Real-time rates and availability Discovery systems cannot reliably recommend bookable inventory from static pages. Google Vacation Rentals depends on live rate and availability feeds [26]
Google Vacation Rentals connectivity Provides an existing demand surface with no referral fee. Free booking links and the approved partner ecosystem [26][27]
Machine-readable property data Bedrooms, beds, location, images, amenities, reviews, policies, and licensing need a shared representation. Google's VacationRental specification models exactly these attributes [28]
High-conversion checkout Avoiding commission is worthless if a larger share of prospects abandon. Airbnb's scale funds continuous conversion investment; 148.3 million Nights and Seats in Q2 2026 [7]
Payments, fraud, and dispute infrastructure Direct booking inherits responsibilities the OTA normally absorbs. Airbnb fee and AirCover documentation against processor pricing that covers payments only [2][4][15]
Portable reputation and trust An unfamiliar host site lacks the marketplace's review corpus and brand halo. Strong stated willingness to pay more for better-reviewed lodging [17]
Attribution A host needs to know whether a booking came from Google, AI, social, paid, referral, email, or a returning guest. The Airbnb pilot itself prices traffic differently by source [1]
CRM and repeat demand First-party acquisition creates the option to retain demand rather than reacquire it every stay. Airbnb prohibits using guest contact information for marketing lists [3]
Content and search distribution Properties need to capture destination, amenity, event, and use-case intent, not only brand-name searches. Search remains a major research channel even as AI grows [16][21]
AI crawler accessibility Answer engines can only cite what they can crawl. ChatGPT Search eligibility requires allowing OAI-SearchBot and its published IP ranges, with placement never guaranteed [41][42]
Social and creator distribution Inspiration frequently occurs upstream of accommodation search. Social appears in travel discovery at rates varying by definition [17][18]
Aggregated demand This is what separates infrastructure from an actual OTA competitor. Airbnb is gaining share despite the wide availability of booking websites [45]
The strategic hurdle is unambiguous. Checkout is the easy layer. A demand network is the hard one, and it is the only layer that would make the rest of the argument true. Haven's zero-commission model is interesting precisely because it forces the question of what else can be monetized: subscriptions, payments, advertising tools, premium distribution, services, and usage-based infrastructure. The research cannot evaluate the viability of that model without Haven's own operating data. It belongs in any eventual article as company strategy, clearly labeled, and never as industry fact. ## 12. SEO opportunity No paid keyword-volume database was used for this research, and no volume figures should be invented to fill the gap. The Airbnb pilot is days old, which makes historical volume particularly uninformative. What follows is based on query intent, topical relevance, SERP structure, and the terminology appearing in the authoritative sources. The opportunity is a pairing: a live news hook attached to an evergreen thesis.
Keyword or topic Intent Priority Role
Airbnb direct booking link Understand the new feature Highest now Primary topical target
Airbnb direct booking links Same, plural Highest now Variant
Airbnb 6% host fee Understand pilot economics High Specific long tail
Airbnb 10% direct booking fee Understand rate variation High Fresh long tail
Airbnb host fees Fee research High, competitive Supporting evergreen section
Airbnb 15.5% host fee Understand the current fee model High Supporting
direct booking vs Airbnb Comparison, commercial High Evergreen thesis
how to get direct bookings Host strategy High Internal link cluster
vacation rental direct booking Category education High Evergreen
Google Vacation Rentals Distribution research and setup High Core differentiation
Google Vacation Rentals direct booking Direct distribution intent Strong long tail Best Haven fit
how to list a vacation rental on Google Implementation Strong Separate cluster
AI vacation rental search Emerging discovery Emerging GEO section
reduce Airbnb fees Cost-sensitive host High commercial intent Natural secondary
Airbnb alternative for hosts Alternative, commercial High but broad Secondary only
"Future of short-term rental booking" works as a thought-leadership concept but is too abstract to serve as the sole primary target. The live Airbnb event is more concrete and more linkable. ### Recommended H1 > Airbnb's Direct-Booking Test and the Future of Vacation Rental Booking It ties the news to the thesis without making a prediction in the title. Alternatives, in rough order of click-through strength: 1. Airbnb Is Testing Lower Fees for Host-Sourced Bookings. That's a Bigger Deal Than It Sounds 2. Airbnb's 6% to 10% Direct-Link Test Reveals What Hosts Are Really Paying For 3. What Happens When the Host, Not Airbnb, Brings the Guest? 4. Airbnb's Direct-Booking Experiment Points to a New Model for Vacation Rental Distribution 5. The Future of Vacation Rental Booking: Airbnb, Google, AI, and Host-Owned Demand The first two should perform better on initial click-through. The last is the more durable evergreen title. ### Suggested article architecture The article should be considerably tighter than this report. An opening answer block of 100 to 150 words covering what Airbnb is testing, the reported 6% to 10% range against 15.5%, and the fact that the booking remains on Airbnb. **What the experiment actually is.** Where demand originates and where the transaction occurs. **The interesting part is not the link. It is the price of demand.** Introduce the acquisition versus transaction distinction. **What 15.5% actually buys.** Demand, trust, payments, support, reviews, protection, conversion. **Direct booking is not free, and that matters.** The $1,000 arithmetic, with acquisition cost named as the missing variable. **New front doors to a vacation rental.** Google Search, Maps, Travel, Vacation Rentals, social. **AI could make property discovery far more specific.** Phocuswright adoption data and Google's August 27 AI Mode launch. **But Airbnb is gaining share right now.** Key Data, placed prominently rather than buried. **The future is probably not OTA versus direct.** A distribution portfolio in which the host owns a growing share of demand. **What the next generation of direct-booking infrastructure has to do.** Haven enters here, and only here. That sequence lets the product appear after the economic argument has been established on its own terms. ### Narrative arc The strongest available frame: > Airbnb has accidentally published a price for the difference between distribution and transaction. Open with the experiment. A conventional Airbnb booking costs most hosts 15.5%. A guest the host brings can now, for selected participants, cost substantially less while still running on Airbnb's infrastructure. [1][2] Then ask what the missing percentage was paying for. The answer is largely the machinery of creating and converting demand, not the mechanics of collecting a card number. Then widen the frame. Google operates an open vacation-rental distribution layer that charges nothing for referrals. AI has become part of travel research at remarkable speed. Social feeds generate inspiration upstream of search. Travelers traverse many pages and platforms before booking. Discovery is fragmenting. [16][19][26] Then interrupt the bullish argument deliberately. Direct reservation share is falling in the best available U.S. dataset while Airbnb's is rising. [45] That interruption is what converts a promotional piece into an argument: > The opportunity is real. The industry has not solved independent demand generation yet. Haven then appears as an attempt to build the missing layer, not as evidence the shift has already occurred. ## 13. GEO opportunity The most reliable generative-engine strategy here is good information architecture and machine accessibility rather than speculative tactics. OpenAI states that eligibility for ChatGPT Search requires allowing OAI-SearchBot to crawl the site and confirming the host or CDN permits traffic from OpenAI's published searchbot IP addresses, and that ChatGPT ranks results using multiple factors with placement not guaranteed. [41][42] Google similarly states that structured data provides explicit clues about a page's meaning, while its structured data guidelines state plainly that Google does not guarantee structured data will appear in search results even when correctly marked up. [29][30] Everything below follows from established indexing principles rather than from theories about how generative systems might behave.
Recommendation Rationale
Place a two to three sentence direct answer immediately below the introduction Makes the central fact extractable without stripping its context
Include a dated fact table covering what Airbnb is testing Lets machines and readers separate verified pilot facts from interpretation
Label established fact, interpretation, and prediction where it matters Prevents the thesis from being repeated elsewhere as reporting
Cite Airbnb's own policy and Help Center pages beside the relevant facts Improves traceability and editorial credibility
Include the $1,000 fee comparison An original, quotable calculation grounded in published rates
Give every statistic a scope, a date, and a source "50% of reservations in Key Data's Q1 2026 professional-manager panel" is far safer than "Airbnb has 50% market share"
Define OTA, direct booking, host-sourced booking, customer acquisition cost, and free booking link once each Establishes explicit entities and their relationships
Keep a dedicated section on arguments against the thesis Makes the article more useful and more citable as a source
Prefer tables to buried prose for comparative facts Improves both human scanning and machine extraction
Maintain published and last-updated timestamps Particularly important while the Airbnb pilot is unstable
For the article itself, `Article` or `BlogPosting` markup with `BreadcrumbList` and coherent organization and author entities is appropriate; both types are documented in Google's structured data gallery. [31][32] For Haven property pages, `VacationRental` is strategically relevant but gated, and adding the markup alone does not produce participation. [28] An FAQ section is worth including because travelers and hosts genuinely ask these questions, not because FAQ markup functions as a lever. - What is Airbnb's direct booking link experiment? - How much does Airbnb charge when a guest uses a host's direct link? - Does an Airbnb direct booking link actually book outside Airbnb? - What is Airbnb's standard host service fee? - Can hosts send Airbnb guests to their own booking website? - How does a vacation rental appear in Google Vacation Rentals? - Does Google charge commission on vacation rental free booking links? - Are direct bookings really commission-free? - Can travelers discover individual vacation rentals through AI? - Will Google or AI replace Airbnb? The honest answer to the last question is channel diversification, not displacement. Answering it any other way would undercut the credibility the rest of the article is working to build. ## 14. Claims we can safely make
Claim Classification Evidence Confidence Qualification
Airbnb is testing host-shared links carrying a lower service fee. Established fact, single source Skift, August 29, 2026 [1] High Limited pilot; Skift is the only published source and worked from host-shared messages
Reported pilot rates range from 6% to 10%. Established reporting Skift [1] Medium to high Skift relays this as reported, not as Airbnb-stated; segmentation is undisclosed
Pilot reservations still occur on Airbnb. Established fact Skift [1] Very high "Direct booking" is misleading without this qualifier
Most hosts on Airbnb's single-fee structure pay 15.5%. Established fact Airbnb Help Center [2] Very high Remaining hosts typically 14% to 16%; 16% in Brazil and Mexico
The pilot can reasonably be read as Airbnb pricing host-originated demand differently. Supported interpretation Pilot structure against standard fees [1][2] High Do not infer Airbnb's underlying costs from promotional pricing
A direct booking avoids OTA commission but still incurs other costs. Established economics Stripe pricing alongside OTA fee structures [2][15] Very high Acquisition, software, and support costs can exceed payment cost
Google can route vacation-rental travelers to a partner's own booking site through free booking links. Established fact Google Hotel Center [26] Very high Integration eligibility applies
Google states there are no fees for Google-generated referrals or bookings from those links. Established fact Google Hotel Center [26] Very high Does not make direct acquisition free overall
AI has become a meaningful travel discovery channel. Established trend Phocuswright, Deloitte [19][21][24][25] Very high Adoption levels vary substantially by survey definition
Google now supports hotel booking inside AI Mode with ten launch partners. Established fact Google, August 27, 2026 [33] Very high Hotels first; no equivalent short-term rental integration exists
Direct bookings can carry disproportionate revenue relative to reservation count. Established for one dataset Key Data Q1 2026 [45] Medium to high Measures revenue concentration, not margin or acquisition cost
Airbnb gained U.S. reservation share while direct lost share in Key Data's Q1 2026 panel. Established for one dataset Key Data via The Host Report [45] Medium to high Self-selected professional-manager panel; single public source
There is no useful universal Airbnb occupancy rate. Supported methodological conclusion Airbnb publishes no bookable-night denominator; AirDNA's own range runs 41% to 67.5% by month [8][48][49] Very high Market-specific occupancy remains measurable and useful
Hosts may come to treat OTAs as one acquisition channel among several. Supported interpretation Hotel precedent, multi-touch discovery, Google and AI infrastructure [16][26][43] Medium to high Direction is plausible; pace and magnitude are unknown
## 15. Claims we should avoid
Tempting wording Why it should not be published
"Airbnb just launched direct bookings." It is a limited pilot, and the booking remains on Airbnb. [1]
"Airbnb now charges 6% for direct bookings." Reported rates range to 10%, and eligibility is undisclosed. [1]
"Airbnb confirmed the program." Airbnb has issued no statement and published nothing about it. [1][5]
"Airbnb's actual transaction cost is 6%." Promotional pricing is not cost accounting.
"Hosts can use these links to take bookings off Airbnb." False for the documented pilot. [1]
"Airbnb is helping hosts build a direct-booking business." The program supplies discounted Airbnb checkout for host-supplied traffic, which is not the same as host-owned acquisition or a host-owned guest relationship. [1][3]
"Airbnb plans to roll this out to all hosts." No evidence exists either way.
"Direct bookings have no fees." Payment, acquisition, and operating costs remain. [15]
"Direct booking is always more profitable." Profitability depends on acquisition cost, conversion, software, and service.
"The average Airbnb occupancy rate is X%." No official metric exists and third-party methodologies differ substantially. [48][49]
"Direct booking is taking share from Airbnb." The best available U.S. data shows the reverse. [45]
"Google replaced ecommerce marketplaces and will do the same to Airbnb." The premise is wrong. Google's commerce surfaces grew alongside marketplaces. [35][36]
"Adding VacationRental schema gets a property into Google Vacation Rentals." Google gates the feature behind Hotel Center access and an Early Adopters Program. [28]
"AI will favor independent rental sites over OTAs." Google's first AI Mode booking partners are large chains and OTAs. The direction is unresolved. [33]
"SEO or GEO can replace Airbnb demand." Nothing in the evidence supports a claim at that scale.
"Haven already rivals Airbnb occupancy with no commission." This is product ambition, not a researched result.
## Closing assessment The research supports a position both more ambitious and more defensible than "hosts should book direct." The likely future is not Airbnb's disappearance. It is the gradual separation of discovery, demand generation, and transaction infrastructure into layers that no longer have to be bought as a bundle. Google's free booking links, the rapid normalization of AI travel interfaces, the durability of hotel-direct distribution, and Airbnb's own experiment in pricing host-sourced demand all point the same way. [1][26][33][43] What has not happened matters just as much. Direct booking has not solved discovery at anything approaching OTA scale, and the most recent U.S. channel data shows Airbnb consolidating rather than losing ground. [45] Which leaves the thesis worth building a company and an article around: > The next major direct-booking company will not win by giving hosts a cheaper place to take reservations. It will win only if it solves the problem Airbnb still solves best: getting the right property in front of the right traveler, generating enough trust to convert that traveler, and doing it repeatedly at an acquisition cost below the commission it displaces. The evidence supports building toward that future. It does not support announcing that the industry has arrived. ## How we know Findings are classified using four labels: **established fact**, **supported interpretation**, **early signal**, and **prediction**. Every numbered citation in this report was checked against the live source on August 30, 2026. Where an earlier research draft misstated a figure, a date, a report title, or a study's methodology, the correction has been applied in the body. Figures that could not be confirmed against an accessible source were removed rather than softened. Three categories of evidence carry unusual weight and should be handled carefully in anything published from this research. Airbnb's pilot is single-sourced. Key Data's channel-mix figures reach the public through one trade outlet and describe a self-selected panel rather than a market census. Phocuswright's AI adoption series is authoritative but its headline percentages shift between waves and between definitions, so any one number needs its wave attached. ### Verification notes and changes from the prior draft The following corrections were applied on August 30, 2026 after checking every cited claim against its live source. **Removed.** A citation to an Airbnb Community forum post reproducing a pilot invitation email with a 6% fee. The post could not be located. The 6% figure now rests solely on Skift's reporting, and the report says so. **Removed.** AirDNA monthly occupancy figures of 67.4% for July 2025 and 58.6% for August 2025, together with the chart built from them. Both AirDNA reports exist and the August report confirms an occupancy dip, but the specific decimals could not be verified because the article bodies render client-side and no secondary source cites them. AirDNA's own published U.S. average of 54.3% and seasonal range of 41% to 67.5% now carry the same methodological point on firmer ground. **Corrected.** The Key Data figures come from an edition labeled Q2 2026 that reports finalized Q1 2026 data. The edition labeled Q1 2026 reports Q4 2025 and shows materially different numbers. **Corrected.** Phocuswright's hotel distribution report is U.S. Hotel & Lodging Travel Market Essentials 2026, published June 2026, and the 52/48 split applies within online bookings only. **Corrected.** The 51% AI click-through figure belongs to Phocuswright's March 2026 research, not July, and its scope is travelers using AI within search engines rather than all AI users. **Corrected.** Expedia's Path to Purchase is two studies, not one tracked panel, and reports U.S. travelers viewing "up to 277" pages rather than an average of 277. Its data dates from spring 2023. **Corrected.** Deloitte's 54% accommodation-research figure applies to the 24% of holiday travelers using generative AI, roughly 13% of holiday travelers overall, and ranks third behind activities and destinations. **Corrected.** Phocuswright's "recognized brands" factor is 32%, third behind price comparisons at 44% and summarized reviews at 33%. **Corrected.** Google's Shopping Graph now holds more than 60 billion product listings, not "tens of billions." The characterization of AI Mode recommendations "linking out to retailers" was removed; Google describes agentic checkout on the merchant's site. **Corrected.** Google opened Shopping to free listings on April 21, 2020, not "April or May." **Corrected.** Booking Holdings' 10-K reports a single consolidated marketing expense line of $8.186 billion. The performance-versus-brand split could not be confirmed from the accessible filing text, so the split characterization was dropped. **Corrected.** The Italian action against Booking.com is an AGCM unfair commercial practices investigation centered on the Preferred Partners programme, not a general antitrust probe. **Corrected.** Citations for Booking.com's commission structure now point to its partner help documentation rather than the "How We Work" page, which blocks retrieval. **Corrected.** The no-guarantee language for structured data appears in Google's general structured data guidelines, not the introduction page. OpenAI's "placement is not guaranteed" language appears in its help center article, not its developer bots documentation. **Added.** Airbnb's Brazil and Mexico fee carve-outs, the requirement that PMS-connected hosts use the single-fee structure, Vrbo's Australia, Japan, and New Zealand exception, and the fact that Airbnb's Nights and Seats Booked metric now combines stays with services and experiences. **Unverifiable and flagged.** Phocuswright does not publicly disclose the sample size behind its 56% AI adoption figure. Phocuswright's own research updates report both 56% and 58% for closely related measures, from different waves. ## Sources - [Skift — Airbnb Is Testing Lower Fees for Hosts Who Bring Their Own Guests (August 29, 2026)](https://skift.com/2026/08/29/airbnb-is-testing-lower-fees-for-hosts-who-bring-their-own-guests/) - [Airbnb Help Center — Airbnb service fees](https://www.airbnb.com/help/article/1857/airbnb-service-fees) - [Airbnb — Off-Platform and Fee Transparency Policy](https://www.airbnb.com/help/article/2799) - [Airbnb — AirCover for Hosts](https://www.airbnb.com/aircover-for-hosts) - [Airbnb Newsroom](https://news.airbnb.com/) - [Airbnb, Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1559720/000155972026000004/0001559720-26-000004-index.htm) - [Airbnb — Q2 2026 Shareholder Letter (August 6, 2026)](https://s26.q4cdn.com/656283129/files/doc_financials/2026/q2/Airbnb-Q2-2026-Shareholder-Letter.pdf) - [Airbnb, Inc. Form 10-Q for the period ended June 30, 2026](https://www.sec.gov/Archives/edgar/data/0001559720/000155972026000027/abnb-20260630.htm) - [Vrbo Help Center — How is the booking fee calculated](https://help.vrbo.com/articles/How-is-the-booking-fee-calculated) - [Booking.com Partner Help — Understanding our commission](https://partner.booking.com/en-us/help/commission-invoices-tax/commission/understanding-our-commission) - [Booking.com Partner Help — Preferred Partner Programme](https://partner.booking.com/en-us/help/growing-your-business/increase-revenue/all-you-need-know-about-preferred-partner-program) - [Booking Holdings Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1075531/000107553126000009/bkng-20251231.htm) - [Reuters — Italy's antitrust agency probes Booking.com (April 22, 2026)](https://www.reuters.com/legal/litigation/italys-antitrust-agency-probes-bookingcom-alleged-unfair-commercial-practices-2026-04-22/) - [Expedia Group, Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1324424/000132442426000008/expe-20251231.htm) - [Stripe — U.S. pricing](https://stripe.com/pricing) - [Expedia Group — Path to Purchase (July 25, 2023)](https://www.expedia.com/newsroom/eg-path-to-purchase-research) - [Expedia Group — 2025 Traveler Value Index (May 20, 2025)](https://www.expedia.com/newsroom/travel-priorities-reinvented-expedia-groups-2025-traveler-value-index-signals-a-shift-in-consumer-priorities/) - [Phocuswright — 26 data-driven insights (2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/26-data-driven-insights-to-make-your-organization-smarter) - [PhocusWire — Shift in travel behavior: AI surge (March 24, 2026)](https://www.phocuswire.com/news/online/shift-travel-behavior-ai-surge-phocuswright-research) - [Phocuswright — The fastest shift in travel behavior just became the default (March 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/The-fastest-shift-in-travel-behavior-just-became-the-default) - [PhocusWire — AI surge: U.S. behavioral shift in travel (July 28, 2026)](https://www.phocuswire.com/ai-surge-us-behavioral-shift-travel-phocuswright-research-2026) - [Phocuswright — The AI surge hits its stride (July 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/the-ai-sure-hits-its-stride-5-takeaways-for-travel-leaders) - [TravelDailyNews — AI adoption reshapes travel planning behavior (March 30, 2026)](https://www.traveldailynews.com/statistics-trends/ai-adoption-reshapes-travel-planning-behavior-in-the-u-s-phocuswright-reports/) - [Deloitte — 2025 summer travel survey (May 20, 2025)](https://www.deloitte.com/us/en/about/press-room/deloitte-americans-plan-to-travel-more-this-summer-but-trips-may-be-less-extensive.html) - [Deloitte — Holiday travel intent results (November 12, 2025)](https://www.deloitte.com/us/en/about/press-room/deloitte-announces-holiday-travel-intent-results.html) - [Google Hotel Center — About vacation rentals on Google](https://support.google.com/hotelprices/answer/10062327) - [Google Hotel Center — Google's vacation rentals partners](https://support.google.com/hotelprices/answer/11946834) - [Google Search Central — Vacation rental structured data](https://developers.google.com/search/docs/appearance/structured-data/vacation-rental) - [Google Search Central — Understand how structured data works](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data) - [Google Search Central — General structured data guidelines](https://developers.google.com/search/docs/appearance/structured-data/sd-policies) - [Google Search Central — Article structured data](https://developers.google.com/search/docs/appearance/structured-data/article) - [Google Search Central — Breadcrumb structured data](https://developers.google.com/search/docs/appearance/structured-data/breadcrumb) - [Google — 3 new ways to plan and book travel in Search (August 27, 2026)](https://blog.google/products-and-platforms/products/search/book-travel-ai-mode/) - [Google — New ways to plan travel with AI in Search (November 17, 2025)](https://blog.google/products-and-platforms/products/search/agentic-plans-booking-travel-canvas-ai-mode/) - [Google — It's now free to sell on Google (April 21, 2020)](https://blog.google/products-and-platforms/products/shopping/its-now-free-to-sell-on-google/) - [Google — Bringing free retail listings to Google Search (June 29, 2020)](https://blog.google/products/shopping/bringing-free-retail-listings-google-search/) - [Google — Buy on Google is now open and commission-free (July 23, 2020)](https://blog.google/products-and-platforms/products/shopping/buy-on-google-is-zero-commission/) - [Google — Google Shopping universal cart (May 19, 2026)](https://blog.google/products-and-platforms/products/shopping/google-shopping-cart/) - [Google Search Central — Product structured data](https://developers.google.com/search/docs/appearance/structured-data/product) - [Google Merchant Center — Product data specification](https://support.google.com/merchants/answer/7052112) - [OpenAI Help Center — Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [OpenAI — Bots developer documentation](https://developers.openai.com/api/docs/bots) - [PhocusWire — U.S. Hotel & Lodging Travel Market Essentials 2026 (June 30, 2026)](https://www.phocuswire.com/news/online/us-hotel-lodging-travel-market-essentials-phocuswright-research-2026) - [Phocuswright — U.K. travel distribution has found its equilibrium (May 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/UK-travel-distribution-has-found-its-equilibrium-Will-it-stay-that-way) - [The Host Report — Airbnb hits 50% of U.S. reservations as direct bookings slide (April 24, 2026)](https://www.thehostreport.com/news/airbnb-hits-50-of-u-s-reservations-as-direct-bookings-slide) - [Key Data — U.S. Key Data Index Q1 2026 for Property Managers](https://info.keydatadashboard.com/hubfs/Key%20Data%20Index/Q1%202026%20U.S.%20Key%20Data%20Index%20for%20PMs.pdf) - [Key Data — EnterpriseData](https://www.keydata.co/products/enterprisedata) - [AirDNA Help Center — How does AirDNA calculate occupancy rate?](https://help.airdna.co/en/articles/8062178-how-does-airdna-calculate-occupancy-rate) - [AirDNA — What's the average occupancy rate on Airbnb?](https://www.airdna.co/blog/average-occupancy-rate-airbnb) - [AirDNA — U.S. Review July 2025](https://www.airdna.co/blog/us-review-july-2025) - [AirDNA — U.S. Review August 2025](https://www.airdna.co/blog/us-review-august-2025) ## How we know Findings are classified using four labels: **established fact**, **supported interpretation**, **early signal**, and **prediction**. Every numbered citation in this report was checked against the live source on August 30, 2026. Where an earlier research draft misstated a figure, a date, a report title, or a study's methodology, the correction has been applied in the body. Figures that could not be confirmed against an accessible source were removed rather than softened. Three categories of evidence carry unusual weight and should be handled carefully in anything published from this research. Airbnb's pilot is single-sourced. Key Data's channel-mix figures reach the public through one trade outlet and describe a self-selected panel rather than a market census. Phocuswright's AI adoption series is authoritative but its headline percentages shift between waves and between definitions, so any one number needs its wave attached. ### Verification notes and changes from the prior draft The following corrections were applied on August 30, 2026 after checking every cited claim against its live source. **Removed.** A citation to an Airbnb Community forum post reproducing a pilot invitation email with a 6% fee. The post could not be located. The 6% figure now rests solely on Skift's reporting, and the report says so. **Removed.** AirDNA monthly occupancy figures of 67.4% for July 2025 and 58.6% for August 2025, together with the chart built from them. Both AirDNA reports exist and the August report confirms an occupancy dip, but the specific decimals could not be verified because the article bodies render client-side and no secondary source cites them. AirDNA's own published U.S. average of 54.3% and seasonal range of 41% to 67.5% now carry the same methodological point on firmer ground. **Corrected.** The Key Data figures come from an edition labeled Q2 2026 that reports finalized Q1 2026 data. The edition labeled Q1 2026 reports Q4 2025 and shows materially different numbers. **Corrected.** Phocuswright's hotel distribution report is U.S. Hotel & Lodging Travel Market Essentials 2026, published June 2026, and the 52/48 split applies within online bookings only. **Corrected.** The 51% AI click-through figure belongs to Phocuswright's March 2026 research, not July, and its scope is travelers using AI within search engines rather than all AI users. **Corrected.** Expedia's Path to Purchase is two studies, not one tracked panel, and reports U.S. travelers viewing "up to 277" pages rather than an average of 277. Its data dates from spring 2023. **Corrected.** Deloitte's 54% accommodation-research figure applies to the 24% of holiday travelers using generative AI, roughly 13% of holiday travelers overall, and ranks third behind activities and destinations. **Corrected.** Phocuswright's "recognized brands" factor is 32%, third behind price comparisons at 44% and summarized reviews at 33%. **Corrected.** Google's Shopping Graph now holds more than 60 billion product listings, not "tens of billions." The characterization of AI Mode recommendations "linking out to retailers" was removed; Google describes agentic checkout on the merchant's site. **Corrected.** Google opened Shopping to free listings on April 21, 2020, not "April or May." **Corrected.** Booking Holdings' 10-K reports a single consolidated marketing expense line of $8.186 billion. The performance-versus-brand split could not be confirmed from the accessible filing text, so the split characterization was dropped. **Corrected.** The Italian action against Booking.com is an AGCM unfair commercial practices investigation centered on the Preferred Partners programme, not a general antitrust probe. **Corrected.** Citations for Booking.com's commission structure now point to its partner help documentation rather than the "How We Work" page, which blocks retrieval. **Corrected.** The no-guarantee language for structured data appears in Google's general structured data guidelines, not the introduction page. OpenAI's "placement is not guaranteed" language appears in its help center article, not its developer bots documentation. **Added.** Airbnb's Brazil and Mexico fee carve-outs, the requirement that PMS-connected hosts use the single-fee structure, Vrbo's Australia, Japan, and New Zealand exception, and the fact that Airbnb's Nights and Seats Booked metric now combines stays with services and experiences. **Unverifiable and flagged.** Phocuswright does not publicly disclose the sample size behind its 56% AI adoption figure. Phocuswright's own research updates report both 56% and 58% for closely related measures, from different waves. ## Sources - [Skift — Airbnb Is Testing Lower Fees for Hosts Who Bring Their Own Guests (August 29, 2026)](https://skift.com/2026/08/29/airbnb-is-testing-lower-fees-for-hosts-who-bring-their-own-guests/) - [Airbnb Help Center — Airbnb service fees](https://www.airbnb.com/help/article/1857/airbnb-service-fees) - [Airbnb — Off-Platform and Fee Transparency Policy](https://www.airbnb.com/help/article/2799) - [Airbnb — AirCover for Hosts](https://www.airbnb.com/aircover-for-hosts) - [Airbnb Newsroom](https://news.airbnb.com/) - [Airbnb, Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1559720/000155972026000004/0001559720-26-000004-index.htm) - [Airbnb — Q2 2026 Shareholder Letter (August 6, 2026)](https://s26.q4cdn.com/656283129/files/doc_financials/2026/q2/Airbnb-Q2-2026-Shareholder-Letter.pdf) - [Airbnb, Inc. Form 10-Q for the period ended June 30, 2026](https://www.sec.gov/Archives/edgar/data/0001559720/000155972026000027/abnb-20260630.htm) - [Vrbo Help Center — How is the booking fee calculated](https://help.vrbo.com/articles/How-is-the-booking-fee-calculated) - [Booking.com Partner Help — Understanding our commission](https://partner.booking.com/en-us/help/commission-invoices-tax/commission/understanding-our-commission) - [Booking.com Partner Help — Preferred Partner Programme](https://partner.booking.com/en-us/help/growing-your-business/increase-revenue/all-you-need-know-about-preferred-partner-program) - [Booking Holdings Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1075531/000107553126000009/bkng-20251231.htm) - [Reuters — Italy's antitrust agency probes Booking.com (April 22, 2026)](https://www.reuters.com/legal/litigation/italys-antitrust-agency-probes-bookingcom-alleged-unfair-commercial-practices-2026-04-22/) - [Expedia Group, Inc. Form 10-K for the year ended December 31, 2025](https://www.sec.gov/Archives/edgar/data/1324424/000132442426000008/expe-20251231.htm) - [Stripe — U.S. pricing](https://stripe.com/pricing) - [Expedia Group — Path to Purchase (July 25, 2023)](https://www.expedia.com/newsroom/eg-path-to-purchase-research) - [Expedia Group — 2025 Traveler Value Index (May 20, 2025)](https://www.expedia.com/newsroom/travel-priorities-reinvented-expedia-groups-2025-traveler-value-index-signals-a-shift-in-consumer-priorities/) - [Phocuswright — 26 data-driven insights (2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/26-data-driven-insights-to-make-your-organization-smarter) - [PhocusWire — Shift in travel behavior: AI surge (March 24, 2026)](https://www.phocuswire.com/news/online/shift-travel-behavior-ai-surge-phocuswright-research) - [Phocuswright — The fastest shift in travel behavior just became the default (March 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/The-fastest-shift-in-travel-behavior-just-became-the-default) - [PhocusWire — AI surge: U.S. behavioral shift in travel (July 28, 2026)](https://www.phocuswire.com/ai-surge-us-behavioral-shift-travel-phocuswright-research-2026) - [Phocuswright — The AI surge hits its stride (July 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/the-ai-sure-hits-its-stride-5-takeaways-for-travel-leaders) - [TravelDailyNews — AI adoption reshapes travel planning behavior (March 30, 2026)](https://www.traveldailynews.com/statistics-trends/ai-adoption-reshapes-travel-planning-behavior-in-the-u-s-phocuswright-reports/) - [Deloitte — 2025 summer travel survey (May 20, 2025)](https://www.deloitte.com/us/en/about/press-room/deloitte-americans-plan-to-travel-more-this-summer-but-trips-may-be-less-extensive.html) - [Deloitte — Holiday travel intent results (November 12, 2025)](https://www.deloitte.com/us/en/about/press-room/deloitte-announces-holiday-travel-intent-results.html) - [Google Hotel Center — About vacation rentals on Google](https://support.google.com/hotelprices/answer/10062327) - [Google Hotel Center — Google's vacation rentals partners](https://support.google.com/hotelprices/answer/11946834) - [Google Search Central — Vacation rental structured data](https://developers.google.com/search/docs/appearance/structured-data/vacation-rental) - [Google Search Central — Understand how structured data works](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data) - [Google Search Central — General structured data guidelines](https://developers.google.com/search/docs/appearance/structured-data/sd-policies) - [Google Search Central — Article structured data](https://developers.google.com/search/docs/appearance/structured-data/article) - [Google Search Central — Breadcrumb structured data](https://developers.google.com/search/docs/appearance/structured-data/breadcrumb) - [Google — 3 new ways to plan and book travel in Search (August 27, 2026)](https://blog.google/products-and-platforms/products/search/book-travel-ai-mode/) - [Google — New ways to plan travel with AI in Search (November 17, 2025)](https://blog.google/products-and-platforms/products/search/agentic-plans-booking-travel-canvas-ai-mode/) - [Google — It's now free to sell on Google (April 21, 2020)](https://blog.google/products-and-platforms/products/shopping/its-now-free-to-sell-on-google/) - [Google — Bringing free retail listings to Google Search (June 29, 2020)](https://blog.google/products/shopping/bringing-free-retail-listings-google-search/) - [Google — Buy on Google is now open and commission-free (July 23, 2020)](https://blog.google/products-and-platforms/products/shopping/buy-on-google-is-zero-commission/) - [Google — Google Shopping universal cart (May 19, 2026)](https://blog.google/products-and-platforms/products/shopping/google-shopping-cart/) - [Google Search Central — Product structured data](https://developers.google.com/search/docs/appearance/structured-data/product) - [Google Merchant Center — Product data specification](https://support.google.com/merchants/answer/7052112) - [OpenAI Help Center — Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [OpenAI — Bots developer documentation](https://developers.openai.com/api/docs/bots) - [PhocusWire — U.S. Hotel & Lodging Travel Market Essentials 2026 (June 30, 2026)](https://www.phocuswire.com/news/online/us-hotel-lodging-travel-market-essentials-phocuswright-research-2026) - [Phocuswright — U.K. travel distribution has found its equilibrium (May 2026)](https://www.phocuswright.com/Travel-Research/Research-Updates/2026/UK-travel-distribution-has-found-its-equilibrium-Will-it-stay-that-way) - [The Host Report — Airbnb hits 50% of U.S. reservations as direct bookings slide (April 24, 2026)](https://www.thehostreport.com/news/airbnb-hits-50-of-u-s-reservations-as-direct-bookings-slide) - [Key Data — U.S. Key Data Index Q1 2026 for Property Managers](https://info.keydatadashboard.com/hubfs/Key%20Data%20Index/Q1%202026%20U.S.%20Key%20Data%20Index%20for%20PMs.pdf) - [Key Data — EnterpriseData](https://www.keydata.co/products/enterprisedata) - [AirDNA Help Center — How does AirDNA calculate occupancy rate?](https://help.airdna.co/en/articles/8062178-how-does-airdna-calculate-occupancy-rate) - [AirDNA — What's the average occupancy rate on Airbnb?](https://www.airdna.co/blog/average-occupancy-rate-airbnb) - [AirDNA — U.S. Review July 2025](https://www.airdna.co/blog/us-review-july-2025) - [AirDNA — U.S. Review August 2025](https://www.airdna.co/blog/us-review-august-2025)