492 answers from ChatGPT, Claude and Gemini, asked on 2026-10-14 the way people ask and about the top businesses of the index. 89% of the answers link a business's own site or its booking software; 29% link a marketplace, most often Uber Eats (14%).
of answers link the business's own site or its booking software
link a marketplace
link Uber Eats, the marketplace named most
answers: 42 to the questions people ask, 450 about businesses
Of the 42 answers to the questions people ask: 100% link a business's own site or its booking software, 17% a marketplace (Uber Eats most); of the 450 about businesses: 88% link a business's own site or its booking software, 30% a marketplace (Uber Eats most).
Updated 2026-10-15. The Lisbon index · Places named, question by question · Markdown · JSON · Share image · All cities.
Each row is a place an answer can send people. The share is the share of all 492 answers that link it; the columns per assistant count that assistant's answers.
| Destination | Kind | Answers | Share | ChatGPT | Claude | Gemini |
|---|---|---|---|---|---|---|
| The businesses' own sites | own site | 425 | 86% | 131 80% | 138 84% | 156 95% |
| Uber Eats | marketplace | 69 | 14% | 16 10% | 15 9% | 38 23% |
| TheFork | marketplace | 53 | 11% | 6 4% | 13 8% | 34 21% |
| Press, guides and forums | press and guides | 40 | 8% | 1 1% | 1 1% | 38 23% |
| Glovo | marketplace | 17 | 3% | 5 3% | 3 2% | 9 5% |
| directory | 15 | 3% | 0 | 2 1% | 13 8% | |
| directory | 12 | 2% | 0 | 0 | 12 7% | |
| Bolt Food | marketplace | 10 | 2% | 0 | 1 1% | 9 5% |
| Restaurant Guru | directory | 9 | 2% | 1 1% | 3 2% | 5 3% |
| Eatbu.com | vendor, direct | 7 | 1% | 3 2% | 2 1% | 2 1% |
| Happycow.net | directory | 5 | 1% | 0 | 0 | 5 3% |
| SevenRooms | vendor, direct | 5 | 1% | 3 2% | 0 | 2 1% |
| Mygon.com | marketplace | 4 | 1% | 1 1% | 1 1% | 2 1% |
| OpenTable | marketplace | 4 | 1% | 1 1% | 3 2% | 0 |
| Resy | marketplace | 4 | 1% | 0 | 0 | 4 2% |
41 more destinations with fewer answers each are in the JSON.
Each assistant was asked the way people ask, with web search on, for up to three places and a link to book or order with each one directly. Of the places it named, how many came with a link to the place's own site, a marketplace, a directory, or no link.
| Assistant | Own site or direct | Marketplace | Directory | No link | Places |
|---|---|---|---|---|---|
| ChatGPT | 100% | 0% | 0% | 0% | 41 |
| Claude | 82% | 0% | 5% | 13% | 38 |
| Gemini | 88% | 0% | 5% | 7% | 41 |
Each assistant was asked, with web search on, for the website, where to book and where to order, for the top-ranked businesses of the index. Share of answers by where they pointed.
| Assistant | Own site or vendor | Marketplace | Directory | Nowhere | Answers |
|---|---|---|---|---|---|
| ChatGPT | 81% | 15% | 1% | 3% | 150 |
| Claude | 85% | 13% | 1% | 1% | 150 |
| Gemini | 93% | 6% | 0% | 1% | 150 |
Asked on 2026-10-14. A place in this index links to its report; a place with a site of its own that is not in the index yet links to a scan; "no link" means the assistant named it but gave no link to its site. Each question links to the full list of places named.
The unit is an answer: the latest answer to each question from each assistant, asked with web search on. The questions are the ones people ask ("best pizza delivered right now", "book me a table for Saturday") and three questions about each of the top-ranked businesses of the index (the official website, where to book, where to order). Every link in an answer is read; a destination counts once per answer that links to it, whether as the place to book or as a source the assistant cited, so the shares say "answers that link this destination" and do not add up to 100%.
Destinations: the businesses' own sites are one row, whatever the business; a booking or ordering vendor on the business's own domain counts as direct; marketplaces and directories are named; listing, review and map sites (Tripadvisor, Google Maps, Yelp) are directories even where the scanner files them with the marketplaces; newspapers, magazines, guides, forums and tourism sites are one row; what is left is listed by registered domain, and a model classifies those unknown hosts once each into directory, press, marketplace, vendor, business or other, with the reason shown. A fragment that looks like a host because a space is missing ("them.here") is dropped. Business names are matched against the index; a place that is not in the index links to a scan of its site when the assistant gave one.
Updated 2026-10-15. Open data: the JSON twin of this page carries every destination. Methodology.
