Local business visibility in AI travel recommendations
Which hotels, restaurants and attractions AI assistants recommend to visitors, how concentrated those recommendations are, and how often the assistants agree.
When a traveler asks an AI assistant where to stay, eat or go in a destination, the answer names specific businesses. This guide shows, from TakeScout's repeated checks across tracked destinations, which kinds of places get named, how few of them take most of the recommendations, and how differently the assistants answer.
Which local businesses do AI assistants recommend to visitors?
For each traveler question, AI assistants converge on a few places: in 97% of the lists we track, one place is named in at least half the answers, out of a median of 47 places named. Which places lead is specific to each destination and each question.
For each traveler question, the place AI assistants name first tends to dominate: the table shows, per category, how many of a list's answers named its first place, and how many places were named in all.
In Québec City, for example: Plains of Abraham in 12 of 12 answers to "Must-See Attractions"; Dufferin Terrace in 10 of 12 answers to "Best Winter Experiences"; Fairmont Le Château Frontenac in 12 of 12 answers to "Best Luxury Hotels"; and Légende in 12 of 12 answers to "Best Restaurants".
| Category | Lists | Median share of answers naming the first place | Lists where the first place is named in half the answers or more | Median places named |
|---|---|---|---|---|
| Attractions | 744 | 100% | 100% (742) | 47 |
| Experiences | 372 | 75% | 96% (356) | 73 |
| Hotels | 620 | 83% | 96% (595) | 31 |
| Restaurants | 620 | 75% | 96% (593) | 51 |
Basis: 28,272 answers in 2,356 lists (one per destination and traveler question) · checks run September 28 to October 14, 2026 · ChatGPT, Claude, Gemini, and Perplexity. Sampled AI answers, not traveler behavior.
What it means for a destination marketer
For a member business, the useful question is which traveler questions it is named first for, and in how many of the answers, not only whether it appears.
Québec City: top picks by question · Traveler intents across the network
Are smaller communities and lesser-known places overlooked by AI?
In our data, recommendations concentrate: five cities took 42% of the city recommendations on broad regional and national questions, and 30 of 60 tracked cities were never recommended there. Within a destination, the three places named most often take 17–31% of a list's mentions (median by category).
When travelers ask broad regional or national questions, 174 of the 413 recommendations of a tracked city (42%) went to five cities: Québec City, New York City, Montreal, Vancouver, and New Orleans. 30 of the 60 tracked cities were never recommended in those answers, including Abbotsford, Atlanta, and Brandon.
Within a destination, the three places named most often take a median share of a list's mentions of 23% for attractions, 17% for experiences, 31% for hotels, and 22% for restaurants.
Basis: 2,940 answers to regional and national questions about 60 cities (each answer counted for each city it could recommend) · collected October 1 to October 14, 2026 · ChatGPT, Claude, Gemini, and Perplexity; 28,272 answers to destination questions (within-destination figures) · checks run September 28 to October 14, 2026 · ChatGPT, Claude, Gemini, and Perplexity. Sampled AI answers, not traveler behavior.
What it means for a destination marketer
For a smaller community, absence from broad questions is the norm in this data, so the relevant measure is the specific traveler questions where it does come up, and which places are named first there.
Presence rate in the methodology · Browse tracked destinations
Do ChatGPT, Gemini, Claude and Perplexity recommend the same places?
Mostly not: in our data, every assistant that answered named the same first place for 11% of questions about a destination. Agreement is highest for day trips (19%) and lowest for restaurants (5.9%).
On 7,437 questions about a single destination, every assistant that answered named the same first place in 11% (809); at least two of them, but not all, named the same place in 59%; each named a different place in 30%.
A question here is one wording of one traveler question for one destination, put to each assistant: 7,437 of them had a clear first pick from at least three assistants. The ranked lists in the other sections pool a traveler question's wordings and assistants into one list per destination (2,480 lists of up to 12 answers, 2,356 of them about hotels, restaurants, attractions and experiences).
Across all their top-three picks, 14,624 of the 21,934 businesses (67%) were named by only one assistant. Basis: 29,760 answers in the closed-businesses report's window · collected October 1 to October 14, 2026 · ChatGPT, Claude, Gemini, and Perplexity.
| Category | Questions | All named the same first place | Some agreed | All different |
|---|---|---|---|---|
| Attractions | 2,231 | 18% | 62% | 19% |
| Day trips | 372 | 19% | 64% | 17% |
| Experiences | 1,116 | 6.5% | 51% | 43% |
| Hotels | 1,858 | 8.0% | 61% | 31% |
| Restaurants | 1,860 | 5.9% | 57% | 37% |
Basis: 29,633 answers with a clear first pick · collected October 1 to October 14, 2026 · ChatGPT, Claude, Gemini, and Perplexity. Sampled AI answers, not traveler behavior.
What it means for a destination marketer
Visibility in one assistant says little about the others, so a business or destination checked in only one of them sees part of the picture.
Model agreement across the network · Model agreement in the methodology
How these figures are produced
Every tracked destination is asked the same traveler questions in several wordings, by every tracked AI assistant, on a recurring schedule; each answer is analyzed for the places it recommends and in what order. Figures here are counts of those sampled answers, not traveler behavior. Full TakeScout methodology.
Last updated October 14, 2026. Figures are recomputed from TakeScout's latest checks when the page loads. More questions: AI Visibility Questions.