Sample. We drew 300 independent restaurants, cafes, bakeries, pubs and bars at random (150 in Vancouver, 150 in Surrey) from OpenStreetMap listings that include a website address, after removing chains, delivery apps and social media pages. 237 of the 300 sites could be read by an automated check; the other 63 blocked the check, were offline or had no readable homepage and are excluded.
Check. Each site was scanned with the same signals as our free AI and local visibility check: the homepage, robots.txt, llms.txt and sitemap.xml were read once, politely, with an identifying user agent. Nothing was stored except the pass, partial or fail result for each signal. No site is named or can be identified from the published data. The scan ran on 23 October 2026.
Score. Each of the 17 signals carries a weight (for example business markup 14 points, AI crawler access 12, FAQ content 10); the score is the earned share of the total, out of 100.
Limits, stated plainly. The sample covers sites that list a website in OpenStreetMap, so restaurants with no website, or whose listing has none, are not represented. We read the homepage only, so a restaurant with strong markup on inner pages could score lower here. The score measures how readable a site is to search engines and AI assistants. It does not measure rankings, traffic, how often an AI recommends a restaurant, or food quality. Results are a snapshot, not a forecast.
Sampling frame. Business listings: © OpenStreetMap contributors, available under the Open Database Licence.