On October 30 we asked ChatGPT where to book a vacation rental in 84 US markets, five times per market. In 61% of the answers, it told the guest to start with a local company. In 3%, it said to start with Airbnb or Vrbo.
That’s 255 answers to 14. If you’ve spent years trying to pull guests away from the OTAs, read that twice.
Now the catch. ChatGPT sends guests to “a local company” far more reliably than it sends them to yours. In markets with four or more established managers, the typical one shows up in 18% of AI answers about its own market. The distance between “book local” and “book with you” is the AI citation gap, and it’s what this study measures.
The Findings In 60 Seconds
- Local beats the OTAs when guests ask where to book. ChatGPT led with a local company in 61% of answers and with Airbnb or Vrbo in 3% (420 answers, 84 markets).
- One word flips it. Ask for “the best website” instead, and an OTA leads 64% of the time, almost always Vrbo.
- AI links to managers far more than Google does. For “[market] vacation rentals,” local managers hold 25% of Google’s top 20 and 58% of ChatGPT’s links.
- Most managers appear. Few appear reliably. 89% of the 132 managers we tracked showed up at least once, but the median manager appeared in 48% of answers, and 18% in crowded markets.
- Size doesn’t buy it. Portfolio size had no meaningful relationship with AI visibility (rank correlation -0.04).
- Ranking sites write the shortlist. ChatGPT read or linked a manager-ranking site in 418 of 420 “best companies” answers. Where two or more of our managers share a market, those on three or more ranking sites were named in 63% of those answers. Those on two or fewer: 23%.
- Those lists mostly run on OTA data. Your Airbnb reviews may now help decide whether AI recommends you as the alternative to Airbnb.
What We Tested
We started with two lists most managers already know: the Comparent 100 (the largest US managers) and AirDNA’s Best Property Managers top 100 (the best-reviewed). We kept companies whose own website names one lead destination, which left 132 managers in 84 markets. National brands and franchises are out because they have no single home market to test.
Each market got three prompts, written the way a guest asks: “[market] vacation rentals,” “best vacation rental companies in [market]” and “where should I book a vacation rental in [market].” We ran them through ChatGPT with web search (five runs each), Google AI Mode, and Google AI Overviews on desktop and mobile. That produced 2,050 AI answers, including 252 we ran to test other wordings. A manager counts as visible if an answer names it or links to it. The full method is at the end.
ChatGPT Tells Guests To Start Local
The typical answer tells the guest to “start with locally managed rental companies rather than relying exclusively on Airbnb or Vrbo.” Another 20% recommend local companies and the OTAs side by side, with no order. Only 3% lead with an OTA.
Airbnb and Vrbo don’t disappear. ChatGPT mentions each in about 95% of these answers, usually as the thing to compare against. But the order matters, because the order is the advice.
When ChatGPT explains itself, the most common reason is selection (35% of local-first openings), then fees or price (24%), then local knowledge and local support (11% each). Plenty of manager websites lead with “book direct and save.” ChatGPT leads with what you have that the OTAs don’t.
This held in most places. Local-first was the majority answer in 53 of 84 markets. In 9 markets, mostly cities and large islands such as San Diego, Savannah and Kauai, ChatGPT read “where” as which neighborhood and never reached the booking question. Orlando was the one market where it usually led with an OTA.
One Word Flips The Answer
Change the question to “what’s the best website to book a vacation rental in [market]” and the result reverses. An OTA led 64% of answers and a local company led 32%. Of those 54 OTA-first answers, 49 started with Vrbo and 5 with Airbnb.
Name the platforms in the question (“should I use Airbnb or Vrbo”) and ChatGPT stays inside that frame every time: 84 of 84.
My read: ChatGPT isn’t pro-local. It answers the question it’s asked. Guests who ask for advice get advice. Guests who ask for a website get the brand it most associates with whole-home rentals. Both kinds of guest exist, and this data can’t tell you how many of each.
AI Treats Local Managers Better Than Google Does
![Stacked bars of link share by site type for [market] vacation rentals. Local managers hold 25% of Google's top 20, 45% of AI Overview links, 58% of ChatGPT links and 63% of AI Mode links.](https://www.buildupbookings.com/wp-content/uploads/2026/10/02-who-gets-the-link.png.webp)
For the same search, “[market] vacation rentals,” OTAs and listing sites hold 36% of Google’s top 20 and local managers hold 25%. In ChatGPT’s links, that flips to 58% local and 17% OTA. Google’s own AI Mode lands at 63% local.
That matters because 87 of our 132 managers (66%) aren’t in Google’s top 20 for their own market, and these are some of the largest and best-reviewed companies in the country. 74 of those 87 still showed up in AI answers. AI gives them a shot Google doesn’t.
Google rank still helps. Managers in the top 20 appeared in a median 64% of AI answers, against 41% for those outside it.
The Gap: Showing Up Isn’t The Same As Getting Named
89% of the managers appeared at least once. That sounds fine until you look at how often. The median manager appeared in 48% of AI answers about its own market, and only 29% appeared in at least three of every four answers.
The biggest split we found is competition. With one established manager on our lists in a market, the median is 68%. With four or more, it’s 18%. An answer names a handful of companies, and a market like the Outer Banks or Destin has more than a handful of good ones. Demand shows the same pattern: 26% in markets with 5,000 or more monthly searches for “[market] vacation rentals,” against 71% in markets under 1,000.
The answers also move around. When a manager appeared for a ChatGPT prompt, it appeared in all five runs only 50% of the time. One screenshot of ChatGPT naming you proves very little.
What Doesn’t Explain It
| Factor | Group | Median share of AI answers | Managers |
|---|---|---|---|
| Portfolio size | Under 250 units | 50% | 68 |
| 250 to 499 | 41% | 40 | |
| 500 or more | 53% | 24 | |
| Domain Rating | Under 20 | 33% | 33 |
| 20 to 34 | 52% | 50 | |
| 35 or more | 48% | 49 | |
| llms.txt file | Has one | 52% | 46 |
| Doesn’t | 46% | 86 |
Portfolio size had essentially no relationship with AI visibility (rank correlation -0.04), and neither did Domain Rating (0.02), although managers under DR 20 do trail. The typical company under 250 units appeared about as often as the typical company over 500. In this group, more doors didn’t mean more mentions.
Robots.txt isn’t the problem either. We read it for 130 of these sites. Only 3 block GPTBot, and none block OAI-SearchBot or ChatGPT-User, the crawlers that fetch pages for ChatGPT’s search and answers. We didn’t test firewalls, which can block crawlers that robots.txt allows. The 46 sites with an llms.txt file showed no real lift, which fits what we said in our GEO guide: it’s cheap insurance, not a ranking lever.
Ranking Sites Write The Shortlist
So where do the names come from? When we asked ChatGPT for the “best vacation rental companies in [market],” it read or linked a manager-ranking site in 418 of 420 answers. AirROI turned up in 82% and Comparent in 80%. Most of these pages are written for owners shopping for a manager, not for guests. 78% of the answers quote a rating or review count, and 21% credit AirROI by name. This is mainly a “best companies” effect: ranking sites came up in 49% of “where should I book” answers and 21% of plain searches.
Being on those pages goes with being named. We fetched the 828 ranking pages ChatGPT retrieved and checked which managers appeared on them. In markets with at least two of our managers, those listed on three or more ranking sites were named in 63% of “best companies” answers. Those on two or fewer were named in 23%, and 14 of those 22 were never named at all. This is a correlation, not proof, and stronger brands land on more lists. But portfolio size barely differed between the two groups (median 313 units against 272).
Here’s the part worth sitting with. Look at what those lists are built from:
| Site | What it scores |
|---|---|
| AirROI | Airbnb listings and guest ratings |
| Comparent | Owner ratings, with verified and sponsored profiles |
| Airbtics | A host score from guest reviews, Guest Favorite share, occupancy and local experience |
| AirDNA | Guest review scores on Airbnb, Vrbo and Booking.com |
| Manager Review | Airbnb guest reviews |
Four of the five run on OTA data. So when ChatGPT tells a guest to skip Airbnb and book with a local company, the company it picks may well depend on how that company performs on Airbnb. That’s our read of how the lists are built, not something we tested, but it means your OTA reputation likely feeds the recommendation that sends guests around the OTA.
Google’s AI Works Differently
AI Overviews rarely show on the head term. 21% of “[market] vacation rentals” searches triggered one, against 94% of “where should I book” searches. If you only check your main keyword, you’ll miss most of them.
AI Mode leans on Google Business Profiles. 93% of its “best companies” answers showed Google business cards, and 84% of the sources in those answers were those cards. “Vacation home rental agency” was the category on 412 of the 526 cards we could count.
AI Overviews cite social. ChatGPT doesn’t. Social and forum links, mostly Facebook, Instagram, Reddit and YouTube, appeared in 57% of AI Overviews and in 2 of 1,260 ChatGPT answers.
What To Do About It
- Measure it properly. Run all three prompts five times each in ChatGPT with search on, then in Google’s AI Mode. Score it as “named in N of M answers” and log which sites get cited. Our GEO guide has the baseline steps.
- Get on the lists. Look up your market on AirROI, Comparent, Airbtics, AirDNA and Manager Review. If you’re on two or fewer, that’s your first job. Check that each one uses the name guests know you by. Airbtics lets companies verify and update their listing, and Comparent shows verified and sponsored profiles.
- Protect your Airbnb reviews, even if direct is the goal. We didn’t test this directly, but three of the five lists above score managers from Airbnb data. Pulling inventory off Airbnb to push direct bookings may weaken the signal that gets you recommended.
- Fix your Google Business Profile. Check your category, your reviews and your website link. It’s a main input for AI Mode’s answers about your market.
- Give AI something specific to repeat. ChatGPT’s case for local starts with selection. Say in plain sentences how many homes you manage, which ones aren’t on the OTAs, and who answers when something goes wrong. This one is our judgment, not a tested result.
- Keep working on Google. Ranking well still goes with being named. AI is a second shot, not a replacement.
What This Study Doesn’t Prove
- Answers aren’t bookings. We measured what AI says, not what guests do. This data can’t tell you how many guests ask these questions or how many book.
- It’s one day. Everything was pulled on October 30, 2026. AI answers change, so treat these numbers as a snapshot.
- Three AI surfaces. We tested ChatGPT, AI Mode and AI Overviews, not Perplexity, Gemini, Copilot or others.
- Wording matters. We tested four ways of asking where to book, and the answer moved each time. Other wordings would move it again.
- These are strong companies. Our managers are among the largest and best-reviewed in the US, and two of our source lists are also sites ChatGPT reads, so they start with an advantage. A typical manager may well do worse. We didn’t measure that.
- Correlation isn’t cause. Being on ranking sites goes with being named. We can’t say one causes the other.
- Home market is a judgment call. We used the lead destination on each manager’s own homepage. A company could do better in a neighboring town we didn’t test.
How We Ran It
- Managers: 132 from the Comparent 100 and AirDNA’s Best Property Managers 2026 top 100, 3 of them on both lists. 65 list entries were excluded with a written reason: national brands, franchises, hotel brands, sites with no guest booking, and sites with no clear home market.
- AI answers: 1,260 from ChatGPT with web search, 252 more testing three other wordings, 252 from Google AI Mode and 286 AI Overviews from 504 Google searches. All via DataForSEO, US, English, with web search on, pulled October 30, 2026. The ChatGPT model reported was gpt-6.
- Google rank: the top 20 desktop results for “[market] vacation rentals,” US, same day.
- Matching: a manager is named if its exact brand appears in the answer text, with image captions removed. It’s linked if its domain appears in the answer’s links. Three companies with generic names were matched by domain only. Visibility pools all three AI surfaces, so ChatGPT, with the most answers, carries the most weight.
- What the answer recommends: a separate AI model read the opening of all 672 booking-question answers without seeing our rule-based labels. The two agreed on 88%. We read a sample of the disagreements by hand, found the reader usually right, and used its labels. Resampling by market puts the 61% between 53% and 68%.
- Site types: about 2,390 domains were classified with AI help from their names and page titles, and we reviewed the most-cited ones by hand. Treat the shares as close, not exact.
- Second source: Ahrefs, kept separate from the panel. Managers who never appeared in our answers had a median 37 AI citations across the platforms Ahrefs tracks, against 124 for those who did. AirROI had 3,132.
- Disclosure: some of these managers are BuildUp Bookings clients. We report aggregates only and name no manager.
Common Questions
Does ChatGPT recommend booking direct or using Airbnb?
It depends on the question. Asked where to book a vacation rental, ChatGPT led with a local company in 61% of 420 answers and with Airbnb or Vrbo in 3%. Asked for the best website to book, it led with an OTA in 64% of answers, usually Vrbo.
Why doesn’t ChatGPT mention my vacation rental company?
In our data the pattern is competition and third-party lists, not size. Among 132 large and top-rated managers, portfolio size had no meaningful relationship with AI visibility. In markets with four or more established managers, the median manager appeared in 18% of answers. ChatGPT read a manager-ranking site in nearly every “best companies” answer, and managers on three or more of those sites were named far more often.
What sources does ChatGPT use for vacation rental companies?
For “best vacation rental companies in [market],” manager-ranking and short-term rental data sites. AirROI appeared in 82% of answers, Comparent in 80%, Airbtics in 40% and AirDNA in 37%.
Do AI Overviews appear on vacation rental searches?
Not often on the head term. 21% of “[market] vacation rentals” searches showed an AI Overview, against 94% of “where should I book a vacation rental in [market]” searches.
Want to know where you stand in your own market? Talk to BuildUp Bookings.
