Your Best Guest Now Asks a Machine First, and It Has Never Heard of Your Hotel

By Rohit Khanna. Written for the BLLA blog. The traffic, citation, and survey figures here come from May through September 2026 and will be stale by early 2027.

In late August I attended a Profound community event on a rooftop in Chinatown. Profound is an AI marketing company. It shows brands how ChatGPT, Gemini, Perplexity, and Google’s AI Mode describe them, and what to change so the answers improve. I spent a while talking with Nick Lafferty, Profound’s founding marketing engineer. We talked about what his team was building and how they were helping companies find the pages and forum threads the machines rely on when they describe a brand, then improve the pages they control and chase corrections on the ones they do not. Marketing, in his telling, becomes measurable when you can see which sources sit beside the answers. That evidence tells a marketer where the next hour and the next dollar should go.

About a hundred people were there, most doing some version of the same work. Two years ago that title was barely visible. No hotel company I know has hired for it yet. The anxiety about who gets found was old. I recognized it from every pre-opening I have run.

On the train home I thought about a general manager in Manhattan whose hotel I audited this spring. In March his hotel got a feature in a national newspaper. He was proud of it, and he had every right to be. I typed a question into ChatGPT that night: a quiet hotel near Madison Square Park with a good bar. It named four hotels. His was not one of them. The newspaper was not cited either.

Who reads a hotel’s press release now?

Three readers. Your marketing team was built for the first two. The first is a journalist. The second is Google’s crawler, and your SEO vendor has spent a decade learning its habits. The third is an answer engine. It reads your release, your reviews, your pages, and every forum thread that mentions you. Then it decides whether to say your name to a family planning a trip, and it never says which signals carried the decision.

The third reader is already part of the planning for many of your guests. Allianz Partners and Ipsos surveyed 2,001 US adults between March 20 and April 14, 2026. Among Americans planning summer travel, 37 percent were using or planning to use AI to help plan their vacations. SparkToro, working from Similarweb clickstream data, found that 68.01 percent of US Google searches from January through April 2026 ended without a click on any result. That figure was 60.45 percent in 2024.

The third reader also sends traffic. Adobe studied more than eight million visits to US travel sites and found AI referrals up 194 percent in May 2026 against the year before. Those visitors stayed 70 percent longer and had a 41 percent lower bounce rate. They also converted 28 percent less often. Adobe did not establish what caused that gap.

If you run a boutique or independent hotel, read those as two separate facts. A growing share of US Google searches end without a click. Visitors who reach a travel site from an AI answer read more and book less. The machine’s description of your hotel now does work your website used to do.

What changed on August 27?

Google began rolling out hotel booking inside AI Mode on Thursday, August 27, 2026, for English-speaking users in the United States. A guest describes the trip in plain language, compares rooms, and pays with Google Pay inside the chat. The ten launch partners are five hotel groups, Hilton, Marriott, IHG, Wyndham, and Choice, and five booking platforms, Booking.com, Expedia, Hotels.com, Priceline, and Trip.com.

No individual independent hotel is among the ten. Independents can still appear through organic results and outside booking links. Google says the hotel or booking platform is the merchant of record for checkout inside the chat. An independent hotel without its own integration may depend on an integrated booking partner for that step. Google says the ranking is organic and depends on the traveler’s request.

The ten names on Google’s list hold the inventory. Discovery and intent, and the conversion of both into a booking, sit outside the list, and that is the ground an independent can still take. Lafferty’s work is being relevant before the booking step begins.

That ground may stay open for a while. On March 5, 2026, OpenAI walked back Instant Checkout in ChatGPT and refocused the product on discovery, according to Skift. Discovery moved to AI faster than transactions did. A hotel can still win the answer without owning the checkout.

There is a smaller detail hoteliers skipped. Google gives the booking partner what it needs to complete the reservation and nothing from the conversation that led there. You get the reservation. You do not get the reason.

What did three luxury operators say on Thursday?

On September 24, 2026, the International Luxury Hotel Association ran a webinar with Travel Media Group on AI and hotel marketing. None of the three panelists quoted a study. All three had met the third reader inside their own portfolios.

Jason Lee, chief technology officer at Travel Media Group, said the output of generative search “is a recommendation rather than a standard list of links.” The guest’s stated preferences shape what the model recommends. The stay meets that expectation, or it does not. The review that follows feeds the next guest’s answer.

Andrew Ladd, vice president of marketing at Noble House Hotels and Resorts, said his team had tracked review sentiment for years. Now it asks a different question. “In responding to guest feedback, can we weave in brand storytelling elements so that those responses feed into AI search indices?” That is a marketing team writing review responses as source material for a machine.

Priya Chandnani, EVP of commercial strategy at Sage Hospitality Group, said the core challenge for a hotel is “remaining the primary source of truth.” People thought large language models would level the field against the online travel agencies, she said. Instead, a hotel has to earn search authority all over again on a new platform.

Ladd added one caution I had not heard from a marketer before. For business travel, an agent that books without friction is welcome. For a luxury vacation, searching is part of the holiday. I do not know how a hotel designs for that. Neither, he admitted, does he yet.

Who does the third reader trust?

Now the part that changes what the March feature was worth. On June 4, 5W, a public relations firm in New York, updated its audit of AI citation sources. It said plainly that it had not verified the research underneath. One Similarweb sample, reproduced in that audit, covered about 600,000 US ChatGPT citation events from January and February 2026, across every category. The Wall Street Journal, The New York Times, Bloomberg, and the Financial Times were absent from the 20 domains ChatGPT cited most. Wikipedia and Reddit together took about a quarter of the citations. The sample says nothing about hotels in particular.

Then Promptwatch, which tracks AI citations, reported a drop in its own sample. Reddit’s share of ChatGPT Search citations fell from 3.83 percent in late July to 0.52 percent after August 14. It called the number provisional. No hotel should build its visibility around one platform, and the number has to be checked every month.

A model can name a hotel without citing the page that persuaded it. The general manager’s feature still carried value with the first reader. It did not secure visibility with the third, at least for that query. The four hotels ChatGPT named had something his did not. Each carried consistent descriptions across its own site, its reviews, video walkthroughs, and travel forums, all saying similar things about the bar. Whether that is why ChatGPT chose them, I cannot prove.

Chandnani’s team at Sage audits brand content across every channel for exactly this reason. Inconsistencies, she said, pop up whenever you compare channels side by side. Consistent information gives an answer engine more sources to compare.

I was part of the problem. In April I told a different client to chase a placement in a national travel section before fixing the hotel’s own pages. I was wrong about the order. The placement ran in June. In August the hotel’s own pages were still waiting.

What does a marketing team have to change?

Five moves. Each one costs something, and the cost is listed.

Write every release for retrieval. Answer engines can draw on specific passages within a page. Place prices, dates, addresses, and checkable claims directly beneath the headings where a reader would look for them. The cost is copy that reads flat to a journalist, so you may need two versions of every release.

Build a consistent footprint across the sources the third reader checks. That means forum threads answered by a named staff member and a YouTube channel with room walkthroughs. It means a Google Business Profile kept current, and a LinkedIn page that posts checkable facts about the hotel before it posts awards. Check the Wikipedia and Wikidata records and request sourced corrections openly. Do not commission a Wikipedia page. The cost is time from people who currently write for the first reader, and a forum presence that goes wrong goes wrong in public.

Treat every review response as a page the machine will read. This is the Noble House move. A response that names the bar, the room type, the restaurant, and the neighborhood gives the model checkable facts in the hotel’s own words. A response that says “thank you for your kind feedback” gives it nothing. Two costs. A human has to write them, because Lee warned that fully automated responses break brand voice fast. And the FTC Consumer Review Rule bans fake reviews, including AI-generated reviews that falsely claim to describe a customer’s experience. It also bans any incentive tied to a positive or negative review. Penalties reach 53,088 dollars per knowing violation. Write responses. Never write reviews.

Measure what the machines say about you every month. Profound does this for brands. Peec AI and Semrush sell versions of it. Answers change by day and by phrasing, so one monthly score can mislead. Run your own 20 questions as well, by hand.

Reconcile the entity data. Your hotel’s name, address, room count, opening year, and restaurant names must match on your site, Google Business Profile, Wikidata, and the booking platforms. Models cross-check, and a mismatch can produce wrong answers. The cost is an unglamorous afternoon with a spreadsheet, and a fight with the platform about who fixes what.

If you run a hotel with a public relations agency on retainer, ask the agency one question this week. Which of these five moves are they already doing, and which do they consider someone else’s job. The second answer tells you where the gap is.

Whose job is this?

McKinsey and Google surveyed 521 marketers in March 2026 and published the results in June. Nearly 60 percent use AI several times a week. Only 28 percent saw their company pursuing a fundamental rewiring of marketing teams and workflows.

Every move above crosses a department line. The review response belongs to the front office and to marketing. The entity data belongs to revenue and to whoever set up the booking engine in 2019. The monthly prompt test needs someone with time, and time is a human resources decision. Lee described review data now reaching operations for service gaps and ownership for capital planning. Nobody on the panel said who owns the connected picture. IT was not on the panel.

What remains unexplained?

The booking gap. Visitors from an AI answer stay longer and book less often. One reading is that they are earlier in their planning. Another is that they arrive with an answer in hand and leave the moment your page fails to confirm it. I do not know which, and I have not met anyone who does.

The decision data. Google does not pass the AI Mode conversation to the booking partner. A 90-room hotel in Charleston has fewer reservations from which to read about changing demand and less pull with the platform that controls the booking path.

This week, open, ChatGPT and Google AI Mode and ask each of them the five questions your best guest would ask about your city. Ask each question three times, because the answers move. Record the exact prompt, the date, every hotel named, and every source cited. Hand the list to the person who runs your PR and repeat the exercise 30 days later.

Rohit Khanna is a hotel lifer who now spends his evenings typing hotel questions into ChatGPT. He then spent 25 years opening and running luxury hotels across Asia, Europe, and North America. Today he runs BodhiMinds.ai, an independent AI strategy and education advisory. He works with travel and hospitality brands on discoverability, reputation management, and reskilling and upskilling their teams for AI. He lives in New York, audits hotels for a living, and still checks whether the machines get the bar right.

Sources:

Allianz Partners and Ipsos, summer travel survey, 2026: prnewswire.com/news-releases/technology-is-transforming-summer-travel-and-american-travelers-are-all-in-302830680.html

SparkToro, zero-click search study, 2026: sparktoro.com/blog/in-2026-less-than-one-third-of-google-searches-still-send-a-click/

Adobe Digital Insights, AI traffic to travel sites, June 2026: business.adobe.com/blog/adobe-report-ai-traffic-travel-sites-surges-200-percent

Google, hotel booking in AI Mode, August 27, 2026: blog.google/products-and-platforms/products/search/book-travel-ai-mode/

Skift, OpenAI checkout walkback, March 5, 2026: skift.com/2026/03/05/openai-chatgpt-checkout-walkback/

5W, citation source audit, June 4, 2026: 5wpr.com/research/citation-source-audit-q1-2026/

Promptwatch, Reddit citations in ChatGPT, August 2026: promptwatch.com/data/reddit-citations-are-dropping-in-chatgpt

McKinsey and Google, marketing organization survey, June 2026: mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/from-anxiety-to-advantage-a-marketing-organization-that-thrives-with-ai

FTC, Consumer Reviews and Testimonials Rule questions and answers: ftc.gov/business-guidance/resources/consumer-reviews-testimonials-rule-questions-answers

ILHA webinar, The New Luxury Hospitality Playbook, September 24, 2026: ilha.org/events/luxury-hospitality-playbook-ai-brand-strategy-hotel-marketing