Key findings
- Almost every site lets AI in: only 1 of 118 blocks AI crawlers such as GPTBot, OAI-SearchBot or PerplexityBot.
- 34% have missing or incomplete structured data about the business (who, where, hours). 15% have none at all.
- 41% have no single clear main heading on the home page, a basic signal of what the page and the business are about.
- 95% do not mark up any questions and answers as FAQ, and 63% have no llms.txt file.
- Online stores: on 90% of the product pages we could read, AI cannot read ratings or reviews, and on 30% price or stock is incomplete.
- Accountants scored lowest (average 86), law firms highest (average 92). In every sector the gaps were the same: clear facts a machine can read with certainty.
Across all 118 sites
Results by sector
| Sector | Sites read | Average score | Business data gaps | No clear H1 | No FAQ markup | No llms.txt |
|---|---|---|---|---|---|---|
| Dental clinics | 19 | 87 | 47% | 53% | 100% | 74% |
| Law firms | 22 | 92 | 18% | 32% | 91% | 77% |
| Accountants | 17 | 86 | 47% | 53% | 88% | 71% |
| Real estate agencies | 19 | 91 | 26% | 32% | 95% | 68% |
| Online stores | 41 | 91 | 34% | 39% | 98% | 44% |
| All sectors | 118 | 90 | 34% | 41% | 95% | 63% |
Online stores: what AI sees on a product page
For 30 of the 41 stores we could open a product page. Shopping answers in ChatGPT and Google rely on product markup: name, price, stock and reviews.
What this means
These businesses already do well on Google and almost none shut out AI. The gap is not access, it is clarity: the plain facts a machine can read with certainty. AI assistants use exactly these facts when they decide which business to name, so the businesses that fix them first have a head start.
Five fixes, in order
- Add business structured data with name, address, phone and hours. Schema generator
- Give the home page one H1 that says what you do and where.
- Mark up your common questions as FAQ and keep the answers short.
- Stores: add Product markup with price, stock and real reviews.
- Check your site and fix the rest. Free AI search check
How we did it
We searched Google with personalization turned off. For professional services we used four cities (Tel Aviv, Jerusalem, Haifa and Rishon LeZion): "lawyer", "accountant" and "real estate agency" in Hebrew. For stores we used five product searches (clothing, jewelry, cosmetics, home and toys). Dental clinics come from our earlier study of 15 October 2026. We kept only the businesses’ own websites and left out directories, portals, review sites, social networks and international brand sites. That gave 127 sites. Each one went through our free AI search checker, which reads only public files: the home page, robots.txt, sitemap.xml and llms.txt, plus one product page for stores. 118 sites could be read; the rest blocked the automated check or did not load.
Limits
The sample is small and comes from businesses that already rank well, so it likely looks better than average. Results are from two days in October 2026. The checker reads the home page (and one product page for stores) and cannot judge content quality. Google says structured data helps it understand a page but does not guarantee any result. We do not publish business names.
Data and citation
The aggregate table is free to reuse under CC BY 4.0 with credit to Balabot. Download the data (CSV)
To cite: Balabot, "Are Israeli business websites ready for AI search? (2026)", 16 October 2026, balabot.ai.