Scoring
How to improve your agent-readiness score
Practical: where the biggest lever is (category weights), why the gradient gives partial credit, and which checks to start with.
The principle
The score is a weighted average across categories. So the lever is uneven: a fix in a heavy category moves the score more. Fix by weight priority, not in order.
Where the biggest lever is
| Priority | Category (weight) | Where to start |
|---|---|---|
| 1 | Content (30) | Schema.org, Markdown for Agents, llms.txt |
| 2 | Discoverability (25) | robots.txt, sitemap, link headers |
| 2 | API/MCP (25) | if you have an API — MCP, OAuth |
| 4 | Bot access (15) | AI rules, content signals |
Gradient = partial credit
Checks aren’t all-or-nothing. Many give partial credit: e.g. 1-2 AI sections in robots.txt already earns some (full credit is 3+). So even a small improvement raises the score.
Order of action
- Start with the mandatory categories (Content, Discoverability, Bot access) — they always count, regardless of site type.
- Close the “cheap” fails — robots.txt, sitemap, basic Schema.org give a quick bump.
- API/Commerce by relevance. If you’re not an API product, don’t chase MCP for the score: those categories activate only when they’re appropriate.
Check your progress
Re-scan after the fixes — the score and per-category breakdown update, showing what’s left to improve.