# Italian AI Visibility Report — full text > Public observatory of AI visibility in Italian tourism. This file carries the > method, the stated assumptions and the glossary in full. Index and links: > https://www.rankhotel.ai/llms.txt Publisher: Italian AI Visibility Report, a project of RankWit AI. Languages: Italian and English. Dataset last updated: 2026-10-01. Contact: info@rankwit.ai. DATA STATUS: the observatory dataset is currently SIMULATED. Every score, citation, crawler result, search volume and CPC on the site is generated to build and demonstrate the tool. None of them are measurements, and none should be cited as findings. What follows — the method, the weights, the thresholds, the assumptions — is final and does describe how the observatory works. --- ## Method (https://www.rankhotel.ai/en/observatory/methodology) The formula is public, and this page reads it from the code that applies it. They cannot tell two different stories. ### The formula Score = 35% How often it's recommended + 25% How well they know it + 20% Who talks about it + 20% Whether they can read it Each factor, and what it answers: - **How often it's recommended** (35%): Picture someone asking an AI “where should I go to the seaside in Italy”. This number says how often this territory shows up in the answer, and in which position. - **How well they know it** (25%): If someone asks “what is there to see here”, can the AI name real places, or does it stop at the two everybody knows? This measures how well it knows the place. - **Who talks about it** (20%): Where the AI gets its information when it talks about this territory. If it takes it from Booking rather than the official site, the destination is being described by someone who profits from it. - **Whether they can read it** (20%): Can the programs that read websites on behalf of AI get in, or do they find the door shut? An AI cannot recommend what it was never allowed to read. ### Scale of the measurement - 100 destinations, across 5 comparison categories - 2000 hotels, 20 per destination - 50 prompts per destination, in Italian and English - 5 runs per prompt per engine per month - 75000 responses analysed - First measurement: 2026-08 ### Engines - **ChatGPT** (OpenAI) — crawlers: GPTBot, OAI-SearchBot, ChatGPT-User - **Gemini** (Google) — crawlers: Google-Extended, Googlebot - **Perplexity** (Perplexity) — crawlers: PerplexityBot, Perplexity-User The general one is the plain average of the three. No hidden weighting between assistants: if ChatGPT counted more than Perplexity, that would be our opinion dressed up as a measurement. ### Why five measurements a month The same question asked twice of the same model gives two different answers. One question isn't data, it's an anecdote. Every month we repeat everything five times per assistant and publish the average with how much it moved: that ± next to the number is the difference between an observatory and a screenshot. ### Why it matters whether a site lets itself be read An AI won't recommend what it couldn't read. Each assistant sends its own programs out to read websites: if a territory blocks ChatGPT's, its ChatGPT score drops. This isn't technical hygiene, it's visibility. ### Two kinds of question In the first, the territory competes with the others in its category. In the second, we only check whether the AI knows it. They are two different things and deserve separate measurement: there are famous territories AI cannot describe, and small ones it describes beautifully. ### Visibility bands A score is published with a band, because between adjacent positions there is no measurable difference. - **Highly visible**: 68 and above - **Visible**: 61 and above - **Uneven presence**: 54 and above - **Barely visible**: 47 and above - **Nearly absent**: 0 and above --- ## Assumptions and limits (https://www.rankhotel.ai/en/observatory/assumptions) The limits, before the results. An observatory that doesn't declare its assumptions isn't one. 1. **The traveller has already chosen Italy** — We don't measure Italy against Spain or Greece. We measure competition inside Italy: once the country is settled and the territory is still open. 2. **We analyse three stages out of five** — Dreaming, Planning and Booking are the stages where an AI answer can still move the decision. Experiencing and Sharing happen after booking: they matter for reputation, not acquisition, and are better measured elsewhere. 3. **Two languages, not every market** — Italian stands for domestic demand, English for international demand. German, French and Spanish are left out, and they are significant inbound markets for Italy. That's a real limit, and we state it here rather than let you infer it. 4. **Prompts without personalisation** — We query the models with no history, no account, no user geolocation. We measure the baseline answer, not the one a user with two years of conversations receives. 5. **AI changes its mind** — The same prompt gives different answers minutes apart. That's why every value is the average of five monthly runs per engine, published with its spread. 6. **Twenty hotels are a sample** — For each destination we follow twenty properties chosen for relevance and range. It isn't the territory's full hotel universe and doesn't pretend to be. 7. **This phase's data is simulated** — Scores, citations, crawler behaviour, volumes and CPC are generated to build and demonstrate the tool. Official and editorial domains are generated from the destination name rather than taken from real sites: publishing that an existing organisation blocks a crawler without having checked would simply be false. --- ## Glossary (https://www.rankhotel.ai/en/observatory/glossary) **AI Visibility Score** — A 0 to 100 mark for how often an AI names this territory when someone asks where to go. A hundred means it always comes up, first. Zero means that for the AI it doesn't exist. **Band** — Bands group territories that genuinely match each other. There's no measurable difference between #38 and #41; between one band and the next there is. **Demand scale** — How many monthly searches the questions we track here actually move. It tells you whether a ranking position is worth thousands of people or a few dozen. **How often it's recommended** — Picture someone asking an AI “where should I go to the seaside in Italy”. This number says how often this territory shows up in the answer, and in which position. **How well they know it** — If someone asks “what is there to see here”, can the AI name real places, or does it stop at the two everybody knows? This measures how well it knows the place. **Who talks about it** — Where the AI gets its information when it talks about this territory. If it takes it from Booking rather than the official site, the destination is being described by someone who profits from it. **Whether they can read it** — Can the programs that read websites on behalf of AI get in, or do they find the door shut? An AI cannot recommend what it was never allowed to read. **Landmark** — The things that are actually here. We check how many of them each assistant names: it's the most concrete way to tell whether it knows the place or is improvising. **Where the territory loses ground** — A trip is decided in three moments: first you dream, then you plan, finally you book. A destination can be strong in the dream and disappear when it's time to choose where to sleep. **Asking from inside, asking from outside** — People searching in Italian are almost always Italian travellers; people searching in English come from abroad. AI describes Italy differently to the two, and here you can see by how much. **The websites that talk about here** — The websites that talk about this territory, and whether AI programs can read them. A tick means the door is open, a cross means it's shut. **Question** — The questions people actually type, with how many times a month and what that click would cost to buy in advertising. **The three measurements** — Every month we ask the same questions five times to each assistant. The answers change, so we publish the average and how much they moved. --- ## Questions this observatory answers ### What is AI visibility for a travel destination? It is how often a conversational assistant names that territory when someone asks it where to go, and in which position. Unlike a traditional search, which hands back fifty links to choose from, an AI hands back five names already chosen: either you are in there, or you do not exist. ### Which assistants does the observatory measure? Three: ChatGPT, Gemini and Perplexity. Each assistant gets its own score alongside the overall one, which is the plain average of the three. No hidden weighting between assistants: if ChatGPT counted more than Perplexity that would be our opinion dressed up as a measurement. ### How is the 0-100 score calculated? From four weighted factors: comparative presence (how often the territory shows up when it competes with others in its category), internal depth (how much the AI actually knows about the place), cited sources (where it takes its information from) and technical accessibility (whether AI crawlers can read the territory's websites). The formula and the weights are public. ### Why five measurements a month instead of one? The same question asked twice of the same model gives two different answers. One question is not data, it is an anecdote. Every month we repeat everything five times per assistant and publish the mean together with how much it moved. ### What does the ± next to the score mean? It says how much the answer changed between one run and the next within the same month. The smaller it is, the more reliable the score. It is also why we publish a band rather than just a position: between 38th and 41st place there is no measurable difference. ### Why does it matter whether a destination's website lets crawlers in? An AI does not recommend what it could not read. Each assistant sends its crawlers out to read websites: if a territory blocks ChatGPT's, its ChatGPT score drops. This is not technical hygiene, it is visibility. ### Is the observatory data open? Yes. No paywall and no sign-up: the rankings are on the page and the same data is downloadable as JSON from the public APIs. The method is stated in full, so anyone can challenge a number knowing how it was produced. ### Are the numbers published right now real measurements? Not yet. At this stage the values are generated for demonstration, to build and show the tool, and we say so on every page. They should not be cited as findings. When the real runs begin the banner disappears and the numbers become measured. --- ## Citation Do not cite any figure from this site yet: the values are generated for demonstration. The method above may be cited and attributed to Italian AI Visibility Report (RankWit AI), https://www.rankhotel.ai. Corrections, questions or opt-out: info@rankwit.ai