AI search marketing for hotels: put the budget where the answers are
Search returns an answer now, not a list of links. This page shows what we measured across the sites we track and what it changes in a hotel marketing plan, with the cost per question written down.
AI sessions measured across the sites we track, last 30 days.
Quick answer
AI search marketing for hotels is the work of getting a property named inside the answers assistants write, and of measuring what that traffic does once it lands. Across the sites we track, in the 30 days to 4 September 2026, ChatGPT was around 90 percent of the AI channel and more than twenty times the next assistant, Copilot. The live counts are in the panel at the top of this page. In hospitality today, optimising for AI mostly means optimising for one assistant and measuring the other four so you see the day that changes.
Where the traffic is today, assistant by assistant
Those same 30 days, in order: ChatGPT first by a distance, then Copilot, Perplexity, Gemini and Claude. Five assistants, and the four behind add up to barely a tenth of the first one. That concentration is the single most useful fact for a marketing budget, because it says the effort is not five programmes running in parallel, it is one programme with four early warning lights.
Two cautions before anyone builds a plan on those numbers. First, they are our sites, not yours. The same count means something different on a property with 500 monthly visits and on one with 50,000, which is why every AI session in the panel is shown next to its share of total visits. Second, a channel at zero is a measurement suspect before it is a business conclusion. Our own Gemini figure sat at zero for a day in early September because of a bug in how we read its citations, not because Gemini ignored anyone.
The measurement itself is first party. The pixel runs on the hotel website and the booking engine, classifies the incoming session by referrer and campaign source, and never estimates from an external panel. Detail on that channel is in AI traffic tracking.
What a full measurement pass costs
On 3 and 4 September 2026 we ran 57 questions across five surfaces, ChatGPT, Claude, Gemini, Perplexity and Google AI Mode, over 7 domains: 285 measurements, 0 errors, 10.46 USD in total. The cost per question is not remotely flat. Google AI Mode 0.0040 USD, Perplexity 0.0129, Gemini 0.0415, Claude 0.0456, ChatGPT 0.0796.
Read that spread once and a budget writes itself. The cheapest surface costs almost 20 times less per question than the most expensive, and in that same pass it surfaced the measured site in 14 of the 57 questions against 10 for ChatGPT. Those questions cover 7 domains of our own travel network and not one hotel, so what travels is the ranking of the surfaces and not the hit rate. Cheapest and best performing were the same surface, which is not what anyone assumes. That was one pass, and a single reading proves nothing. It is also why a fixed monthly fee for AI visibility tracking tells you nothing until you know how many questions and how many surfaces it buys.
What this changes in the marketing plan
A marketing lead who has these numbers stops arguing about whether AI matters and starts allocating. Three lines of work, in the order that pays first.
Start with the entity, not the page. On local questions, Google AI Mode overwhelmingly cites Google business profiles rather than websites, so the fastest win for a hotel is a complete and correct profile. That case is made with the numbers in hotel AEO.
Then fix what blocks a crawler. A blocked bot makes every other investment worthless, and it takes minutes to check with the robots.txt checker and the llms.txt checker.
Only then write. The content team gets its brief from measured questions rather than from a keyword tool, which is the subject of AI content strategy, and the revenue manager gets the channel on the same board as the rest of demand, which is AI demand signals. The pilar guide, how to appear in ChatGPT, walks the whole sequence.
FAQ
ChatGPT, by a wide margin. In the 30 days to 4 September 2026, across the sites we track, ChatGPT was around 90 percent of the AI channel and more than twenty times the next assistant, Copilot, with Perplexity, Gemini and Claude behind it. The live counts are in the panel at the top of that page.
Our pass of 3 and 4 September 2026 cost 10.46 USD for 57 questions on five surfaces, 285 measurements in total. Per question the range was 0.0040 USD on Google AI Mode to 0.0796 USD on ChatGPT, with Perplexity at 0.0129, Gemini at 0.0415 and Claude at 0.0456.
Five: ChatGPT, Claude, Gemini, Perplexity and Google AI Mode. Every question runs on all five in the same pass, which is what makes the answers comparable, and the 57 questions of the 3 and 4 September 2026 pass produced 285 measurements with 0 errors.
No. Classic SEO competes for a position in a list of links. AI search competes to be written into a single answer, and it fails in different places, such as a business profile that beats your own website on local questions. It also concentrates: 90 percent of the AI sessions we measured in the 30 days to 4 September 2026 came from one assistant.
Weekly is enough for a hotel, and it is affordable: at 10.46 USD per pass, the price we measured on 3 and 4 September 2026, 52 passes come to about 544 USD a year. Daily adds cost without adding decisions, because rate and content changes rarely move an answer inside 24 hours.
As many as it has real demand for, which is rarely a round number. In the 3 and 4 September 2026 pass the 57 questions were shared unevenly across 7 domains of our own travel network, from 1 question on the smallest site to 29 on the largest, because the set is built from each site own Search Console queries and not from a package.