We score the page the way a machine reads it.
Traditional SEO tools grade you for Google's blue links. We grade you for the moment an answer engine decides whether to quote you. The audit is deterministic and parser-based — the same signals every time, no black box — so every point maps to a fix you can actually make.
Six constructs cut across every check.
On top of the seven display categories, checks roll up into cross-cutting constructs. They're how we model trust, not just structure — and YMYL topics raise the bar automatically.
E-E-A-T
Does a real, identifiable author stand behind the page, with evidence an engine can verify?
JTBD
Does the page actually resolve the task the searcher arrived with — directly and early?
YMYL
For health, finance, and safety topics, the trust bar is raised and weighed more heavily.
AI Access & Crawlability
Robots rules, render budget, and payload size decide whether a crawler ever sees the content, before anything else can matter.
Machine Readability
Clean structure and explicit markup let a parser separate the actual content from navigation, ads, and boilerplate.
Citeworthiness
Specific claims, original data, and direct answers give an engine something concrete enough to cite instead of paraphrase.
130+ atomic checks. Growing toward 200.
Each check is one declarative, evidence-backed unit — it runs, returns a status and a snippet, and carries a weight and severity. A representative sample:
| Check ID | Category | Severity | What it verifies |
|---|---|---|---|
| http-status | Technical SEO | Critical | 2xx, no soft-404 or redirect chain |
| robots-allows-ai | AI Accessibility | Critical | GPTBot / PerplexityBot not blocked |
| heading-order | Content Structure | High | Single H1, no skipped levels |
| answerable-lead | Answer Format | High | Extractable answer in the opening |
| author-entity | Authority | High | Named author with linkable identity |
| jsonld-article | Structured Data | Medium | Valid Article / Product JSON-LD |
| faq-block | Answer Format | Medium | Q&A pairs an engine can lift |
| content-depth | Content Quality | Medium | Substantive, original coverage |
The critical gate is a ceiling — never a floor.
Checks that don't apply to your page are excluded from the math entirely — they never silently drag you down. The remaining signals blend by weight into seven category scores and one overall 0–100 with a letter grade.
A failed criticalcheck — say, you're blocking AI crawlers outright — caps your overall score (e.g. at 20) no matter how strong everything else is. But the gate only ever lowers a high score; it never inflates a low one. If a page is genuinely weak, the number stays honest. That's the whole point: presence, not dominance.