Greece: are small-business websites ready for AI?
393 small-business websites built by one website provider, measured in November 2026. What ChatGPT, Google AI and Copilot can read on them.
- Exact trade declared
- 0.5%
- AI-readable FAQ
- 0%
- llms.txt describing the business
- 2.8%
- Auto-generated llms.txt
- 90.6%
Nearly every site has an llms.txt, but it is the builder’s automatic page list: nothing about the business itself.
Public data · no site or provider named
The figures
Greece next to the French study: 38,803 sites built by 9 French providers, measured the same way.
| Signal | Greece n = 393 | France (9 providers) n = 38,803 |
|---|---|---|
| llms.txt describing the businessA summary of the site written for AI, at the root of the domain. A page list generated by the site builder does not count. | 2.8% | 0% |
| Auto-generated llms.txtAn llms.txt generated by the site builder: the list of pages and their descriptions, nothing about the business. | 90.6% | 23.6% |
| Rules for AI crawlersrobots.txt names GPTBot, ClaudeBot, PerplexityBot or another AI crawler. | 0% | 0.1% |
| AI crawlers blockedrobots.txt bars an AI crawler (or every crawler) from the whole site. | 0% | 0% |
| AI-readable FAQQuestions and answers declared as structured data (FAQPage). | 0% | 1% |
| Structured data presentAt least one schema.org JSON-LD block on the homepage. | 95.2% | 72.3% |
| Exact trade declaredA schema.org type that says what the business does (Plumber, Restaurant, Attorney…). | 0.5% | 3.1% |
| Generic type onlyAmong sites with structured data: only “LocalBusiness” or “Organization”, no trade. | 99.5% | 95.8% |
| Empty descriptionThe structured-data description field exists but is empty. | 0% | 2.4% |
| Ratings declaredAn average rating (AggregateRating) in structured data. | 0% | 0.9% |
| Meta descriptionThe summary search engines show under the page title. | 98.7% | 92.1% |
| SitemapA sitemap.xml declared or present. | 99.7% | 98.7% |
| HTTPSThe site answers over a secure connection. | 100% | 100% |
| Mobile-readyThe viewport tag is present. | 80.2% | 100% |
What it changes
AI only recommends what it understands. A site that doesn’t say clearly what trade it is in, where, and for which questions, leaves the answer to competitors who do.
- 01The exact trade (Plumber, Restaurant, Attorney…) lets AI match you to the right questions.
- 02A structured FAQ gives AI answers ready to quote.
- 03An llms.txt that describes the business — not a page list — tells AI what you do and what matters.
Method
When: November 2026, a single measurement at that date.
What is measured: the raw HTML of the homepage, plus /robots.txt, /llms.txt and /sitemap.xml. No JavaScript is run: anything added only by script is not counted.
Sample: small-business websites built by subscription website providers that run on one of two white-label site-builder platforms, identified by their technical signature (hosting, provider credit in the footer or legal notice). Country = the site’s domain extension. Sites with no identifiable provider are left out.
Excluded: unreachable sites, parked or near-empty pages, duplicates after redirects.
Limits: we measure what sites declare, not what AI answers. Each sample reflects its providers’ templates and is not representative of all businesses in the country.
Privacy: aggregate figures only; no site or provider is named.
France: the full French study (in French) covers 9 providers, with breakdowns by trade and region.
What about your site?
I test your business on 10 real customer questions in ChatGPT, Google AI and Copilot. A one-page PDF within 24 hours.