How ScamMinder Scores Websites

What this page is

Every ScamMinder result page shows a trust score from 0 to 100 and a verdict such as Safe, Warning, Scam or Phishing. This page explains, in plain language, where those numbers come from, what they can and cannot tell you, and how to dispute a result. It is the reference behind every website analysis we publish.

What we check

An analysis combines automated checks from independent sources. Each check runs on its own; no single source decides the outcome.

• Domain registration (WHOIS / RDAP): how old the domain is, whether the registrant is hidden, and the registrar.
• DNS and email authentication: hosting provider and country, and whether SPF and DMARC records exist (their absence is common on throwaway scam domains).
• TLS certificate: validity, issuer, certificate type and expiry. A certificate proves encryption, not honesty, so it is a small signal.
• Threat-intelligence feeds: Google Web Risk and PhishTank, plus advisory infrastructure reputation (Spamhaus DROP and ASN-DROP, and ipinfo). A confirmed listing on a phishing or malware feed weighs heavily and can decide the verdict on its own; infrastructure reputation is context and never decides a verdict by itself.
• Popularity and history: Tranco traffic ranking and Wayback Machine history. A site that claims to be established but has no history is a red flag; a site with years of archived history is harder to fake.
• Content review: we load the site's pages and an AI model reviews what the site says and sells, its contact details, pricing claims, urgency tactics, impersonation of known brands, and patterns we have learned from confirmed scams (for example investment "guaranteed return" schemes).

How the score is built

1. AI content review produces a first score and a written assessment with reasons and red flags.
2. Guardrails then adjust that score with deterministic rules that cannot be argued around: a confirmed phishing listing caps the score; a very new domain limits how high it can go; strong legitimacy evidence (established age, working email authentication, consistent identity) can lift a harsh AI score. The guardrail pass runs three times and the final pass is the one we publish.
3. Review disposition decides whether the evidence is complete. If checks were blocked or contradictory, the page says so and the score is marked as needing review rather than presented as certain.

A score of 70 or above is required for a Safe verdict. Scores in the middle band render as Warning: review the findings before you trust the site. Low scores with strong scam or phishing signals render as Scam or Phishing. Inactive means the domain did not respond, and Unknown means we could not gather enough evidence either way.

What the verdict does not mean

• A low score is not an accusation of a crime. It means the signals we could observe resemble those of fraudulent sites, or that we could not verify the site at all.
• A high score is not a guarantee. Scammers can buy old domains and copy legitimate designs. Use the score alongside your own judgement, and never send money on the strength of any single tool.
• Adult, gambling and piracy sites are scored on trust and safety signals like every other site. We label those pages for search engines and do not promote them.

Expert moderation and the verified badge

Some pages carry an expert review. A ScamMinder moderator has examined the site by hand and published a verdict that supersedes the automated one; the automated result stays visible on the page for transparency. Expert moderation is editorial and unpaid — it cannot be bought. Separately, a site owner can pay for Owner Verification, which proves control of the domain and the owner's identity and adds an "Owner Verified" badge for one year. That badge is provenance, not a safety rating, and it is not a place in any listing. Nothing we sell changes a score or a verdict — not Owner Verification, not Expert Analysis, not Get Help, not Premium, not API credits. We sell no score change, no rating removal and no guaranteed outcome.

Disputes are free

If you believe a result is wrong, use the "Was this analysis accurate?" control on the page, the report link below the score, or the form at /dispute. Reports go to a human review queue, and a site owner who publishes the DNS TXT record we issue reaches a reviewer directly. You can also request a fresh automated analysis, which re-runs every check above; a domain can be re-analysed at most once every 24 hours. Expert Analysis is an optional paid service for a private written investigation, and the free dispute path never depends on it.

Community signals and history

Where readers have left approved reviews, rated an analysis accurate or inaccurate, or filed reports that moderators reviewed, the page shows those counts. Pages also show their verdict history: each dated status change from automated reanalysis or expert moderation. Reader feedback never changes a score automatically; it prioritises human review.

Freshness

Each page shows when the site was first and last analysed. Popular pages are reanalysed periodically and on request; a reanalysis replaces the published result and records the change in the verdict history.

Contact

Questions about a result, a takedown request for a page about your own site, or a data correction: use the contact page. Our disclaimer and privacy policy describe the limits of the service and how we handle data.