More and more travellers are skipping the search bar and simply asking an AI assistant where to book. We asked five of them the same 30 questions in June and again in September. In one market the answer changed completely. Here’s what they named, what moved, and why “most recommended” is still not the same as “best.”
Type “recommend me a reliable Turkish tour operator” into ChatGPT, and you’ll get a confident answer in seconds. Ask Gemini to find you a car-rental firm, and you’ll get another confident list. Ask Grok, and you’ll get nine names where Claude might give you three. Discovery is moving from the search engine — where you scroll links and decide for yourself — to the AI assistant, where a single answer arrives pre-sorted. When the assistant names three brands, those three have effectively been placed on a shelf at eye level, and everything else is in the back of the store.
This is the second edition of a quarterly study, so for the first time we can say not just who is on that shelf, but what moved.
Why this is worth paying attention to
Turkey’s travel market is large on both the resident and inbound sides. Domestic tourism spending reached TL 555 billion in 2025, up 32.4% year on year, across nearly 67.9 million domestic trips (TÜİK, 2026, as reported by Hürriyet Daily News). On the inbound side, the country earned $65.2 billion in tourism income from 63.9 million visitors, up 6.8% and 2.7% respectively (TÜİK, 2026, via Anadolu Agency).
Discovery is increasingly digital. IMARC put Turkey’s online travel market at $5.46 billion in 2025, projecting roughly 9% annual growth through 2034 and tying it to mobile-first booking and AI-supported trip planning (IMARC Group, 2026); internet use among 16–74-year-olds reached 90.9% in 2025 (TÜİK, 2025). The shift toward AI at the research stage is visible globally: Adobe reported generative-AI-driven traffic to U.S. travel sites up 3,500% year on year in July 2025, with 29% of surveyed consumers using AI to plan trips (Adobe Analytics, 2025) — a signal that assistants are becoming discovery and comparison interfaces, not just booking engines.
Car rental, the second market in this study, is measured more by fleets than booking revenue: in Q4 2025 daily car rental ran 138,300 vehicles across 1.09 million contracts, and the rental sector as a whole purchased TL 278.1 billion of vehicles in 2025 (TOKKDER/NielsenIQ, 2026; Anadolu Agency). The operator base mixes international chains (Avis/Budget, Enterprise, Europcar, Sixt) with domestic firms (AVEC, Garenta, QCar, Rent Go, Zeplin, Ziraat Filo), and comparison platforms such as Enuygun and Yolcu360 list 100+ rental firms — a digital-discovery layer that maps directly onto what the assistants surface.
All of which makes one question pressing: as travellers move to AI assistants, who do those assistants actually name when someone asks for a tour operator or a rental car? Over a single day in September 2026 we put the same 30 Turkish-language questions we used in June to five leading AI assistants — GPT-5.6-luna-pro, Gemini, Claude, Perplexity and Grok — asking each question five times: 750 answers, 748 of them usable, across two markets.
One market just changed hands
In June, car rental was led by international chains and Garenta sat third. It now leads outright, named in 66.6% of car-rental answers against 39.3% in June — the largest move by any brand in either market. Avis is second, Enterprise third. The chains have not gone anywhere; they simply no longer top the list.
Tour agencies went the other way and closed. Jolly Tur still leads, named in 74.6% of tour answers, and this quarter every single ranked tour agency is Turkish — the foreign column is empty. Across all tour-agency answers, 98.2% of brand mentions go to Turkish firms. Car rental remains genuinely mixed at 62.6%.
Below the leaders, each list still thins out fast. Of 205 brands we tracked, 47 were named often enough to rank at all (15 tour agencies, 32 car-rental firms). Visibility is concentrated: a small group occupies most of the space, and a long tail barely registers.
Tour agencies — the current ranking
Below are the tour agencies that cleared the 5% mention threshold, ranked by AI Visibility Score. This is a visibility ranking, not a quality ranking.
Car rental — the current ranking
A score is a position against this quarter’s leader in that market, so it cannot be compared between the two lists; against June’s it is given in the Δ column as direction rather than exact size. For all 205 tracked brands, the raw mention rates behind each score, and the full methodology, see the current edition below.
The assistants moved too
You might assume five AI systems would converge on the same answer. At the very top they largely do — Jolly Tur appears in every assistant’s most-named list. Everywhere else they diverge, and this quarter they diverged in a new way: the assistants themselves changed.
Three of the five are running the exact same model version we tested in June, and all three now name fewer brands per answer than they did then. Grok is still the most generous, naming around nine brands in a typical answer, while Claude is the most selective at roughly three — but measured against June’s own brand list, Grok dropped from 9.4 to 8.0 and Claude from 4.3 to 2.5. A narrower shelf means a smaller or newer operator has less room to make the cut.
One assistant largely stopped searching
In June, about 98% of all answers were built by searching the live web. This quarter it is 87.57% — and almost the entire drop is one model. Gemini’s search rate fell from 90.7% to 39.3%, and 91 of the 93 answers in the whole run that used no web source at all came from it. The other four searched on essentially every answer.
When the assistants do search, they still lean on brands’ own websites more than any other kind of source — though less than before, at 41.4% of citations against 48% in June. They are reading more places, and more shallowly.
One detail worth pausing on: 8 brands had their own websites cited as sources and were never named in a single answer. Being read is not the same as being recommended.
The most important finding isn’t a ranking
Here’s what we’d most like readers to take away: being recommended by AI does not mean a brand is good. When an assistant names a brand often, it usually means one thing — there is a lot of information available about that brand. It does not mean the brand’s tours are better, its cars newer, its prices lower, or its service more reliable. AI visibility measures how easily a brand is found, not how good it is. Travellers should not read an AI recommendation — or this report — as a verdict on quality.
And because this is now a series, one more caution: a brand’s score moves when the leader moves, not only when the brand does. Two of the five assistants also changed between editions. That is why the Δ column above gives direction rather than exact size. The full report prints every trend figure beside the raw mention rate that underlies it, which is the layer that travels between quarters unchanged.
Read the full report
This article is a summary. The complete report contains both leaderboards in full, the breakdown for each AI assistant, the question-type winners, the Turkish-versus-foreign analysis with its decomposition, the source analysis, the trend against June, and the full methodology. We repeat this study every quarter; you’ll find the current and past editions listed below.
About this study: Herm.io is a consumer-behavior and marketing-data company. We don’t sell SEO or AI-ranking services, no brand paid to appear in the report, and a brand’s score is a measure of visibility only, not an endorsement. To keep the analysis neutral, we also excluded any AI citations to Herm.io’s own pages from the source data. Brands curious about where they sit in the data are welcome to book a call for a neutral walkthrough of the findings; it’s free and there’s nothing to buy.
Mert Can Elkaya
Contributor
I'm a product builder working at the intersection of product, fintech, and growth. From martech and venture capital to leading product at a proptech platform and co-founding a fintech startup, I help teams—and shoppers—make smarter, more confident decisions.
More from Mert