How Trust Scores Work

How Trust Scores Work

A plain-language overview of how Veridex turns the five inputs — a website, an email, a text message, a QR code, or a screenshot — into the 0–100 Trust Score you see on every result page.

What we look at

Five inputs, one set of rules.

Every analyser shares the same deterministic-rule engine, the same weighted signals, and the same three-band score. Pick the input that matches what you want to check.

Websites

Open checker

Pasted domain or URL — no network call from us. We read the host, scheme, path and query string and run the rules against them.

Emails

Open checker

Paste the email headers and body. We parse the sender, scan for pressure language, and fold each link back through the URL rules.

Text messages

Open checker

Paste an SMS or chat. We read the sender shape, scan for scam phrasing, and fold each link back through the URL rules.

QR codes

Open checker

Upload a QR image. We decode the payload on the server, then route it through the URL rules or the SMS-style keyword rules.

Screenshots

Open checker

Upload a screenshot. We extract the text, find any links, and run the keyword and URL rules against what the image says.

The principles

What Veridex is — and isn't.

Behind the score is a small set of commitments about how the system behaves, what it produces, and where its limits sit.

  • Hundreds of technical signals run on every input. Behind every Trust Score is a fixed set of weighted rules — sender domain checks, certificate lookups, lookalike-character detection, urgency-phrasing scans and more. We run the same rules on every input, in the same order, with the same weights.
  • AI explains results in plain English. The screenshot analyzer returns a short paragraph, a confidence number, and three concrete next steps — written in the language a non-technical reader can act on, not a dump of signal ids.
  • Scores can change as new information becomes available. A Trust Score is a snapshot of the signals we saw at the moment of the check. New information about a site, a sender, or a campaign can shift the picture tomorrow — and the same input re-checked can return a different score.
  • No automated system is perfect. False positives and false negatives are part of the deal. The matched signals are surfaced in plain text next to the verdict so you can argue with the call — and the click stays yours.
  • Veridex is a decision-support tool, not a definitive authority. We surface the score, the matched signals, and a concrete next step. We never auto-block a message, never claim something is "verified safe", and never replace your judgement.

What could go wrong

Honest about what the score does not tell you.

  • We can miss a scam. New patterns land ahead of the rules. The score reflects the signals we ran at the moment of the check, not a guarantee.
  • We can call a real thing a scam. Rule weightings err on the side of caution. The matched signals are shown so you can override the call.
  • The score is a snapshot, not a label. Re-running the same input later can return a different score, and a clean score today does not lock in a clean score tomorrow.
  • No inbox pipe, no crawl layer. The check reads what you paste; it is not connected to your mailbox or browsing history.

Read further

For the rule-by-rule walk, and the people behind it.

This page is the short version. The longer walk — every signal category, the band cutoffs, and the AI explanation — lives on the methodology page, and the story of the small crew that runs the rules lives on the about page.

Methodology — signal-by-signalAbout Veridex

Try a verifier

The score shifts when the signals fire.

Run a check on a real input and watch the score update as each rule matches.

Check a URLCheck an emailText message analyzerQR code analyzerAnalyse a screenshot