| Check | Weight | What it looks at |
|---|---|---|
| Crawler access | 20% | Whether AI crawlers can reach the site's pages at all — blocked bots mean invisible pages. |
| Structure | 25% | Headings, sections, lists, and clean HTML that an AI can parse into answers. |
| Metadata | 25% | Titles, descriptions, and structured data — the bits assistants quote directly. |
| Content | 15% | Clear, factual copy that answers the questions the client's customers actually ask. |
| Performance | 5% | Load speed. Slow pages get skipped by crawlers working on a budget. |
| FAQ | ±10 pts | Question-and-answer content, the format AI assistants cite most. An adjustment, not a weight: a weak or missing FAQ takes 10 points off, a strong one adds 10. |
How the checks become the score
Each check produces its own 0–100 score. The five weighted checks together are worth 90 points, so structure (25%) and metadata (25%) move the needle most and performance (5%) least. FAQ then adjusts the total: a weak or missing FAQ costs the site 10 points, a strong one adds 10, and one in between leaves the score unchanged. If crawler access scores zero, the score is capped at 40. Every issue found becomes a ranked fix in the report, ordered by impact.
Note
If one check can't run — say the site's server blocks the scan — it scores zero for that check rather than failing the whole scan. The report says what happened.