Methodology · 8 min read

How Trust Scores Work: Our Methodology Explained

Transparency is core to Scamtrustor. Here is exactly how we calculate Trust Scores, what factors matter most, and why some sites score low.

Our Commitment to Transparency

At Scamtrustor, we believe every Trust Score should be explainable. Unlike black-box AI systems, our scoring engine provides detailed reasons for every score — reasons you can read, question, and understand.

The Six Scoring Pillars

Our Trust Score (0-100) is calculated across six weighted categories:

  • Review Quality (35%): Bayesian-weighted, time-decayed ratings from verified community members
  • Authenticity (20%): Anti-fraud signals, review pattern analysis, behavioral signals
  • Identity (15%): Domain age, WHOIS data, registration patterns, business verification
  • Security (15%): SSL certificates, HTTPS, security headers, DNS health
  • Transparency (10%): Claimed profile, owner responsiveness, public information quality
  • Performance (5%): Uptime, response time, reliability signals

Why Separate Risk Score?

The Trust Score reflects community trust and positive signals. The Risk Score (also 0-100, higher = more dangerous) captures negative fraud signals independently. A site might have a moderate Trust Score but a very high Risk Score — meaning some users trust it, but our systems detect serious danger signals.

Fraud Probability

Our fraud probability (0.0 to 1.0) is calculated from a weighted combination of behavioral fraud signals: review manipulation, infrastructure shared with known scams, sudden reputation changes, and more. A probability above 0.7 triggers automatic review flags and admin notification.