Finding
Can AI tools read contractor websites?
Often only partly. Of 2,635 home services homepages we fetched without running JavaScript, 1,395 (52.9%) had no machine-readable business facts, and 233 (8.8%) showed fewer than 500 characters of text.
What was missing
| Check | Homepages | Share |
|---|---|---|
| No business schema (LocalBusiness, a trade type, ProfessionalService or Organization) | 1,395 | 52.9% |
| No structured data of any kind | 977 | 37.1% |
| Under 500 characters of text before JavaScript | 233 | 8.8% |
| Under 500 characters and no business schema | 199 | 7.6% |
A site with almost no text before JavaScript usually had no business schema either. Those sites give a tool that does not run JavaScript close to nothing to work with.
Crawler blocking was uncommon
Among 2,839 sites whose robots rules we could check (a site with no robots.txt blocks nothing):
| Crawler | Used for | Blocked | Share |
|---|---|---|---|
| OAI-SearchBot | ChatGPT search | 27 | 1.0% |
| ChatGPT-User | ChatGPT, when a person asks it to open a page | 27 | 1.0% |
| PerplexityBot | Perplexity | 31 | 1.1% |
| GPTBot | OpenAI model training | 92 | 3.2% |
| ClaudeBot | Anthropic | 91 | 3.2% |
| Google-Extended | Google AI training controls | 90 | 3.2% |
Read these as minimums. This only covers robots.txt. A firewall or security service can block crawlers without showing up in that file.