One score hides the problem.
Seven dimensions reveal the fix.

The overall score is a weighted average of 7 dimension scores. For a blog post the weights are Answer extractability (25%), AI-friendly structure (20%), Specific & quotable (15%), Evidence & sources (10%), Original information (10%), Topic coverage (10%), and Freshness & authority (10%); weights shift by content type, so a comparison page weights original information higher. Each dimension combines deterministic content analysis (heading structure, date detection, entity counting) with LLM evaluation (semantic quality, answer placement, claim specificity).

Use the score as a directional editing benchmark, not a citation guarantee. Higher scores indicate that more of the measured extractability, structure, evidence, originality, quotability, coverage, and freshness checks are satisfied. The dimension breakdown matters more than chasing a single threshold.

A single score hides the root cause. A page might rank 80 overall but score 35 on Answer extractability because the key definition is buried in paragraph 6. The 7-dimension breakdown tells you exactly which aspect of your page AI models struggle with, so you fix the right thing.

Yes. Citegrade stores every audit result. You can re-run an audit after applying fixes and see the score change per dimension on the report page. Pro and Scale plans include score history and trend tracking.

No. A high citation readiness score means your page has the structural and semantic qualities that make AI citation likely, but LLM behavior depends on training data, query context, and model architecture. Citegrade optimizes the factors you can control.