How ScamCallCheck builds its records

Public agency sources

Complaint files are normalized to canonical phone numbers and stored as daily aggregates by source and complaint subject. Re-importing a source updates that same grain instead of duplicating complaints.

Agency complaints represented

772,956

Numbers with agency data

656,083

Complaint-date coverage

October 1, 2003–September 9, 2026

Latest agency row imported

September 9, 2026

Community reports

Community posts are submitted by visitors describing their own calls or texts. Phone-like strings, email addresses, links, and Social Security number patterns are removed before publication. Repeated text, residual sensitive information, and submission bursts are held for review. Reports can also be flagged or challenged through the review and removal process.

Risk labels

The score combines complaint and report volume, recency, recent velocity, category severity, and the share of posts identifying a legitimate business. Signals use diminishing returns so a large raw count cannot dominate every other factor. A strong legitimate-business consensus lowers the result. Exact internal weights are withheld to make coordinated score manipulation harder.

Search-index quality tiers

Useful number pages are submitted in three separately measurable sitemap groups: pages with published community reports, agency-only pages with at least two corroborating complaints, and pages with one agency complaint from the last 12 months. Empty pages and single stale complaints can still be looked up directly but are not submitted for indexing.

Campaign observations

Campaign pages group recurring language found in published community reports over a rolling 24-month window. They describe message patterns only. Matching wording does not establish that numbers share an operator, brand, or owner, and ScamCallCheck never infers those identities.