Priority actions

1. Error Content without JavaScript
   Almost no content in the raw HTML. Most AI crawlers (GPTBot, ClaudeBot, PerplexityBot and others) do not run JavaScript and therefore see an almost empty page. Use server-side rendering (SSR) or pre-rendering, or make sure the theme/page builder outputs the text in the HTML.
2. Warning Page title (<title>)
   Use a unique, descriptive title of 10–60 characters.
3. Warning Meta description
   Write a 50–160 character description summarising the page.
4. Warning Key schema types
   Add WebSite. It tells AI services who is behind the site.
5. Warning HTTPS
   HSTS (Strict-Transport-Security) tells browsers to always open the site over https. It can be enabled under SSL/TLS Certificates in Plesk.
6. Warning Image dimensions in HTML
   Set width and height on images so the browser reserves space. It stops the page from jumping while loading (CLS in Core Web Vitals).
7. Warning Link text
   Link text tells search engines and AI what the target page is about. Use e.g. "See bathroom prices" instead of "Read more". Icon-only links need an aria-label.
8. Warning Last updated
   Provide dateModified/datePublished in JSON-LD (done automatically by Yoast/Rank Math on posts). AI services prefer content they can see is up to date.

AI bots

The bots for AI search and citations matter most – if they are blocked, the site does not appear in answers. Blocking training bots is often deliberate and does not lower the score.

The server test fetches the page with the bot's identity from this server. Some firewalls block fake bots by IP address while letting genuine bot traffic through – confirm any blocks in the firewall log. Googlebot, Bingbot and Applebot are only checked against robots.txt.