Here's how the analytics on this site are produced, and where to be cautious.
Judgments are downloaded from The National Archives' Find Case Law service as structured LegalDocML/Akoma Ntoso XML, under a computational-analysis licence.
Case and legislation references are taken from the structured enrichment markup in each judgment, then linked into a citation graph. Some historic references cannot be resolved to a specific judgment and are shown by citation only.
How a citing judgment treats a cited case (applied, distinguished, overruled, etc.) is inferred from the citing sentence. The default classifier is rule-based and conservative — it favours precision and labels unclear mentions as a neutral "cited". Treatment labels are indicative; always read the citing judgment.
The influence score is PageRank over the citation network: a case scores highly when influential cases cite it, not merely when it is cited often.
Appeal outcomes (allowed/dismissed/…) are classified automatically from the concluding paragraphs and are indicative, not authoritative.
Coverage is limited to what Find Case Law publishes; automated extraction makes mistakes; and citation counts reflect this dataset, not all of case law. Do not rely on this site for legal decisions — verify against the official judgment.