What we found.
0 of 36 scored a B or better. The best score was a C. The median was 52 out of 100. How the score works
5 of 36 turn AI agents away at the door. Their bot protection refuses an agent that says who it is, so a customer's ChatGPT or Claude never sees a page. How to fix it
20 of 25 multi-location brands we could measure hide their locations inside a map. Their locations page has no plain links to individual locations, so an agent asked to book at one club cannot reach it. Another six chains gave us no locations page we could find from the homepage at all. How to fix it
11 of 22 location pages state no address, phone or hours for AI. Of the location pages we could open, half carry none of the three as structured data. Five carry all three. How to fix it
1 of 31 gives an agent anything to do. Only one site we could reach registers WebMCP actions, and none publish a manifest of what an agent can do there. How to fix it
16 of 18 with customer accounts give no way to let a member's agent in. No OAuth discovery, so the only way for a member's own AI agent to manage their account is the member's password. How to fix it
7 of 31 put a captcha on the page. An agent that has done everything right still cannot submit the form. How to fix it
8 of 31 publish an llms.txt briefing for AI. The cheapest fix on the list, and three in four have not done it. How to fix it
How we measured.
Each brand's main website was scanned once on 2026-10-24 by AgentReadyBot, which identifies itself on every request, obeys robots.txt and never tries to get past a bot wall. A scan reads public pages only: the homepage, its locations page and one location page where there is one, and the files agents look for. Brands are counted, not named. Sites change, so a brand's score today may differ. The full method, check by check, is on the methodology page.
Where does your site land?
The same scan, free, in about a minute.