What we check (and why)
Our score is citation-focused: it weights the signals that actually decide whether you get found and quoted by ChatGPT, Claude, Perplexity and Google's AI Overviews. We're deliberately honest — for example, we tell you JSON-LD earns Google rich results but barely affects AI citations, so it carries little weight here.
| Signal | Weight | Why it matters |
|---|---|---|
| Server-rendered content | 35 | Content must be in the raw HTML to be read & cited — the #1 requirement. |
| Content-Signal | 15 | Governs search / AI-retrieval / AI-training in robots.txt. |
| Markdown for agents | 15 | Clean markdown to agents cuts ~80% of tokens. |
| MCP manifest | 15 | Lets agents act through a defined door. |
| llms.txt | 10 | Optional curated map for agents. |
| JSON-LD | 10 | Google rich results — minimal AI-citation impact. |
Homepage-level checks, point-in-time, HTTP-based — a snapshot, not an audit. For the deeper agent & commerce layer (OAuth, x402/UCP/ACP, agent skills), Cloudflare's Is It Agent Ready is excellent and complementary. To go from score to fix, read the pillar or grab the free AIO Playbook skill.