On 12 November 2026 Aleryc Digital Studio automatically read 733 business websites in Genoa and Umbria, taken from OpenStreetMap: for 683 the server answered the request for the robots.txt file, and 664 home pages could be analysed.
Almost none shut the door on artificial intelligence: only 5% block at least one of its reading programs (35 out of 683). But few introduce themselves: only 12% of the websites say through structured data that they are a local business (79 out of 664), just 4% give their opening hours that way (28 out of 664), and only one home page in five shows phone, address and opening hours together in the text (123 out of 664, 19%).
Only 12% of business websites in Genoa and Umbria say through structured data that they are a local business (79 out of 664 home pages analysed on 12 November 2026; websites taken from OpenStreetMap), according to research by Aleryc Digital Studio.
How to cite this study: Aleryc Digital Studio, “Genoa and Umbria, ready for artificial intelligence?”, 12 November 2026, https://aleryc.com/research-genoa-umbria-businesses-ai-ready-2026 · Data (CSV, CC BY 4.0): https://aleryc.com/assets/ricerca/ai-ready-genoa-umbria-2026-aggregate-data.csv
More and more people ask ChatGPT, Gemini, Claude or Perplexity where to eat, where to stay or whether a shop is open. To answer, these assistants also read business websites. There are two questions: do business websites let their programs in? And when they do, do they clearly say who they are, where they are and when they are open?
On this page you’ll only find overall figures, always shown with the total they are based on. We don’t publish business names or rankings. The method and its limits are explained at the end.
A note on transparency: Aleryc Digital Studio also offers services for visibility on Google and in AI assistants. That is why we publish the method and the limits. Nobody can guarantee an appearance in AI answers, and we don’t promise it.
The numbers at a glance
The door to AI is open
robots.txt is a small file that tells automated programs which pages they may read. 576 websites out of 683 have one (84%). Of the 10 AI programs we checked (those from OpenAI, Anthropic, Perplexity, Google, Apple, Common Crawl and ByteDance), at least one is blocked on 35 websites out of 683 (5%):
- in 27 cases (4%) with a rule that names the program;
- in 8 cases (1%) only with a general ban that applies to everyone.
Not all AI programs do the same thing. Some collect text to train models. Others are used to search for and cite pages when someone asks a question:
- OAI-SearchBot and ChatGPT-User for ChatGPT;
- Claude-SearchBot for Claude;
- PerplexityBot for Perplexity.
At least one of these is blocked on only 13 websites out of 683 (2%). When there is a rule with a name, it almost always concerns the training programs. In 10 of the 27 cases the rule is the one prepared by Cloudflare, a service widely used for website security.
What it means: almost everyone lets AI in, often without knowing it. That’s fine, as long as it’s a choice. Blocking training programs does not remove the website from Google; blocking search programs can remove the business from that assistant’s answers. If you use Cloudflare or a security plugin, it’s worth checking what is switched on.
Few introduce themselves
Structured data is a few lines of code, invisible to visitors, that tell machines what kind of business you are, where you are and when you are open. For businesses with premises there is a dedicated type, LocalBusiness, with subtypes such as Restaurant, Hotel or HairSalon. 79 websites out of 664 (12%) use it. More than one website in three, 247 out of 664 (37%), has no structured data at all.
Websites that do have structured data mostly use the kinds that many website tools create on their own, such as WebSite (329 websites) or Organisation (229). They are useful, but they don’t say where the business is or when it is open.
Among the 79 websites with LocalBusiness:
- the address is almost always there (74, 94%);
- the phone number often (64, 81%);
- opening hours only in 28 (35%);
- the map position (geo) in 31 (39%);
- links to social profiles (sameAs) in 26 (33%).
What it means: it is easier for an assistant to answer with clear, consistent information. Google points out that no special structured data is needed to appear in its AI answers: what counts are the basics, meaning a readable website with correct information that is the same everywhere.
Opening hours are the big gap
Opening hours in structured data appear on only 28 websites out of 664 (4%). In the home page text, a time slot such as “12:00–15:00” or “from 9 to 13” appears on 147 websites out of 664 (22%). Even among restaurants, bars and cafés, where opening hours matter most, it appears on only 72 websites out of 199 (36%). Among shops and beauty salons, on 52 out of 214 (24%).
What it means: “is it open now?” is one of the most common questions. If the website doesn’t say, the assistant looks elsewhere, for example in portal listings or reviews, or doesn’t answer.
Phone yes, opening hours hardly ever
We looked at what can be read in the home page text without opening other pages:
- a phone number appears on 464 websites out of 664 (70%);
- an address, meaning street and number or the area’s postcode, on 468 (70%);
- all three together (phone, address and opening hours) on 123 websites out of 664 (19%);
- none of the three on 136 (20%).
| Category | Phone, address and opening hours on the home page |
|---|---|
| Restaurants, bars and cafés | 32% (63 out of 199) |
| Shops and beauty salons | 20% (42 out of 214) |
| Hotels, farm stays, B&Bs and hostels | 7% (18 out of 251) |
For hotels, farm stays and B&Bs, opening hours matter less: often there is only the check-in time.
12% of home pages (77 out of 664) have fewer than 300 characters of readable text without running the page’s programs (JavaScript). In 20 cases the text only appears afterwards: anyone who reads the page without JavaScript, as almost all AI programs do, finds an almost empty page.
What it means: someone who asks an assistant for “phone and opening hours of…” gets a better answer if the website states them clearly, as text and not inside an image.
Information that doesn’t match
We compared the phone number and postcode written on the home page with those in the OpenStreetMap listing of the same business:
- phone: where both were present, at least one number matches in 261 cases out of 330 (79%). In 69 cases out of 330 (21%) no number matches;
- postcode: it matches in 194 cases out of 211 (92%).
What it means: assistants combine different sources: website, map listings, portals, social media. If the phone number changes from one source to another, the answer may be wrong. A check once a year is enough.
The technical basics are there
- The title for search engines is there on 636 websites out of 664 (96%).
- The meta description on 426 (64%).
- The page language is declared on 602 (91%).
- A site map for search engines (sitemap) is found on 523 (79%).
A curiosity: 86 websites out of 664 (13%) respond at the /llms.txt address, a file intended for AI. It is not a standard: it is a proposal, and Google has said it is not needed for its Search. In at least 20 cases it is a Wix website, where the file may be created by the platform itself.
Genoa, the province of Terni and the rest of Umbria side by side
The measurements are the same in the three areas. Differences of a few points should not be read as significant, especially for the province of Terni, where there are about a hundred websites.
| Measurement | Genoa | Province of Terni | Rest of Umbria |
|---|---|---|---|
| Blocks at least one AI program | 8% (17 out of 224) | 4% (4 out of 103) | 4% (14 out of 356) |
| No structured data | 39% (83 out of 214) | 39% (40 out of 102) | 36% (124 out of 348) |
| Local business structured data | 12% (25 out of 214) | 15% (15 out of 102) | 11% (39 out of 348) |
| Opening hours on the home page | 30% (64 out of 214) | 16% (16 out of 102) | 19% (67 out of 348) |
| Phone, address and opening hours on the home page | 24% (51 out of 214) | 10% (10 out of 102) | 18% (62 out of 348) |
| Phone matching OpenStreetMap | 82% (69 out of 84) | 75% (46 out of 61) | 79% (146 out of 185) |
Local business structured data in Genoa: 12% here (25 out of 214), 11% in the October study (25 out of 232). It is the same measurement on a different total: here only the Genoa home pages we could read on 12 November count, in October the websites analysed at that time.
How we did it
- Source of the websites: they are the websites of our two October studies, Genoa and Terni and Umbria, taken from OpenStreetMap: data © OpenStreetMap contributors, available under the ODbL licence. We took again those that responded in October, 733 in total: 239 in Genoa, 106 in the province of Terni and 388 in the rest of Umbria, from a randomly drawn sample. They are the 724 websites whose home page we had read in October (232 in Genoa, 106 in the province of Terni, 386 in the rest of Umbria) plus the 9 that at the time asked in robots.txt not to be visited (7 in Genoa, 2 in the rest of Umbria). Compared with the 740 websites that worked in October (247 in Genoa, 107 in the province of Terni, 386 in the rest of Umbria), the 16 whose content could not be read at the time are missing (15 in Genoa, 1 in the province of Terni).
- Categories: restaurants, bars and cafés; hotels, farm stays, B&Bs and hostels; shops and beauty salons.
- What we measured: in robots.txt, the rules for 10 AI programs, with the official names checked on 12 November 2026 on the operators’ websites; on the home page, structured data (LocalBusiness with address, phone, opening hours, position and social profiles); in the home page text, phone, address and opening hours; title, description, language, sitemap and llms.txt.
- Rules: on 12 November 2026 we first read robots.txt and then, if allowed, the home page, once only. For sitemap and llms.txt we only requested the headers, without downloading the files. Requests were made a few at a time, with a name that says who we are. We did not run JavaScript: we read the pages as they arrive from the server, as almost all AI programs do.
- The totals: for robots.txt, the 683 websites whose server responded count, that is 733 minus 50: 41 servers closed the connection to our program and we did not insist, 3 domains no longer exist, 3 websites gave certificate errors or refused the connection, and 3 responded with a server error. For the other checks, the 664 home pages read count, that is 733 minus 69: 12 websites skipped because they ask not to be visited, 51 with a network error (the 47 above and 4 cut off while the home page was arriving), 3 blocked by an anti-bot system and 3 that responded with an error. Each percentage shows its total and is rounded to the nearest whole number.
The limits of this research
- OpenStreetMap does not contain every business, and the research only covers websites that responded in October.
- We only looked at the home page: opening hours or an address on a “Contact” page are not counted.
- The measurements come from a single day and a single moment.
- A website can block AI programs in ways that robots.txt doesn’t show, for example through Cloudflare settings. We never presented ourselves as an AI program to find out.
- A rule in robots.txt may have been added by a plugin, the hosting provider or Cloudflare, and not necessarily by the business.
- Phone, address and opening hours are recognised with automated rules: a check-in time counts as opening hours, and a few cases may have been missed or counted by mistake.
- Opening hours were not compared with OpenStreetMap, because the formats are too different for a reliable automated comparison.
- Having structured data and an open door does not guarantee appearing in AI answers: the research measures the basic conditions, not the answers.
What a business can do
- Write your name, address with postcode, phone number and opening hours on the home page, as text and not inside an image. It’s free and you can do it yourself.
- Check that this information is identical on the website, in your Google listing, on social media and on OpenStreetMap.
- If you use Cloudflare or a security plugin, see whether blocking AI programs is switched on, and decide for yourself whether to keep it.
- If you feel up to it, add local business structured data with the same information, and check it with Google’s free Rich Results Test. We cover this in our guide to SEO, AEO and GEO.
- For basic checks on your website there is our guide Does your website work? 7 free checks.
Aleryc Digital Studio is the studio of Alessandra De Luca, web designer: it builds websites and an online presence for local businesses, freelancers and small companies. She works from Genoa, where the business is based.
How to cite this study
You can reuse the figures by citing the source with a link to this page:
Aleryc Digital Studio, “Genoa and Umbria, ready for artificial intelligence?”, 12 November 2026, https://aleryc.com/research-genoa-umbria-businesses-ai-ready-2026
The aggregate data, with totals by area and category, can be downloaded as a CSV file: aggregate data of the study (CSV) (data dictionary). It contains no website addresses or business names. The aggregate data is published under the CC BY 4.0 licence: you can reuse it by citing Aleryc Digital Studio. It contains information from OpenStreetMap, available under the ODbL licence. For questions about the method: info@aleryc.com.