Beat the Scam Scam checks, alerts, and consumer protection guides
Home / Research / Methodology

Search and AI visibility research methodology

A repeatable method for measuring search discovery, AI citations, consolidation effects and click-through rate without overstating what the data can prove.

What we measure

Google Search Console: final web-search clicks, impressions, CTR and average position for the latest complete 28-day period, compared with the immediately preceding equal period. Page and query views are retained internally; the public dataset includes site totals and named focus pages.

Bing AI Performance: total citations, daily cited-page counts, cited URLs and the grounding-query sample supplied by Bing Webmaster Tools. The public report uses the dashboard’s 30-day exports; a three-month export is retained internally for longer comparisons.

Monthly collection process

  1. Export the 30-day and three-month Overview, Pages and Grounding queries CSV files from Bing AI Performance.
  2. Run the measurement script, which retrieves final Search Console data, normalises the Bing exports and creates an immutable dated snapshot.
  3. Check totals, date coverage and focus-page mappings before publishing the normalized public JSON and CSV files.
  4. Compare equal 28-day periods. Use seven-day results only as an early directional check.
  5. For a redirected or consolidated guide, assess the old and new URLs together until the source URL disappears from platform reports.

Interpretation rules

  • A Bing citation is a source appearance, not a ranking, endorsement, visit or guarantee of prominent placement.
  • Bing grounding queries are a sample, and Bing’s generated intent and topic labels can change.
  • Search Console omits anonymized queries; filtering and aggregation can make dimension totals differ.
  • Average position is contextual. We prioritize impressions, clicks and CTR, then use position to diagnose page-level changes.
  • An editorial release is treated as an intervention. A before-and-after change can correlate with the release but does not prove causation.
  • Visibility data for this site is not evidence of national scam prevalence or financial harm.

Privacy and reproducibility

No message submitted to the scam checker, personal information or subscriber data is collected for these reports. Each release names its source periods, publishes normalized files and preserves its raw source exports internally for audit.

Primary platform documentation

Method version 1.0 · Published 18 July 2026. Material changes will be documented on this page.