AI Technical Readiness Scorecard
A 0–100 score for the technical side of AI discovery. We check crawler access, llms.txt, sitemap, schema.org markup, canonical and semantic signals in one fetch, then rank the fixes by how many points they recover.
Access, files, schema
robots.txt rules, llms.txt, sitemap.xml, JSON-LD and canonical signals in one pass.
Weighted, not vague
Each check carries points, and fixes are ordered by the score they recover.
No AI guessing
Pure technical checks against the served response — the same thing a crawler sees.
Checks run against the HTML, headers and root files exactly as a crawler receives them.
Client-rendered pages will score low on content even when they look complete in a browser.
How the 100 points split
Crawler access
Robots rules for Googlebot, OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-SearchBot and Bingbot, weighted by how much each one drives citations. A 5xx robots.txt counts as a full block.
Discovery files
A valid llms.txt, a reachable sitemap.xml, a Sitemap: line in robots.txt, and a robots.txt that answers with 200.
Structured data
JSON-LD present, an Organization or WebSite entity so the brand is identifiable, and a page-level type such as Article, Product, FAQPage or SoftwareApplication.
Semantic signals
A 15–65 character title, a 50–165 character description, exactly one H1, a self-referencing canonical, and Open Graph title plus image.
Delivery
HTTPS, a 200 response and a text/html content type — plus enough text in the raw HTML to be worth citing.
What the grades mean
85–100, Excellent. Assistants can fetch the page, identify the brand and extract a clean summary. Keep the files current.
70–84, Good. Working, with one or two gaps that cost citations — usually missing JSON-LD or Open Graph tags.
50–69, Needs work. Typically a blocked crawler or an unreadable robots.txt. Fix the access pillar before anything else.
Below 50, Critical. The page is either unreachable to crawlers or has no content in the served HTML, so nothing else can compensate.
What this does not check
Rankings, backlinks, model sentiment and whether a real assistant happens to mention you. Those are measured by prompt-level AI visibility tracking, not by a one-off fetch.
Core Web Vitals and Lighthouse performance are also out of scope. This is a discovery audit — can an AI system reach the page, understand it and quote it.
Fixing the gaps
Start with crawler access: a blocked OAI-SearchBot or PerplexityBot makes every other signal irrelevant. The checker explains each decision and offers a corrected robots.txt.
Then publish an llms.txt with the free generator, and add JSON-LD plus a single H1 to the pages you want quoted.