How we know

How we know what we show

Every answer comes with the evidence behind it, from public information only. Here’s how we work, and where we stop.

What you can count on

  • Evidence, not verdicts

    Every answer shows what we found and where, so you can check it yourself.

  • Public information only

    We read what any visitor can see. Nothing behind a sign-in.

  • A polite scanner

    CheckAnySiteBot says who it is, follows each site’s robots.txt and never hammers a site.

    About our scanner
  • Honest about gaps

    When we can’t check something, we say so instead of guessing.

  • Tested on real sites

    We test ourselves on thousands of sites we know, and publish what we get wrong.

    How we test
  • Corrected in the open

    Anyone can report a mistake, and every change we make is in the changelog.

    Changelog

Where each answer comes from

Each tool answers one question from its own sources, and says when it can’t.

  • AI Website Detector
    Where we look
    The traces AI site builders leave in the sites they make.
    When it comes up empty
    Coding assistants and hand edits leave none, so “none found” never means “made by a person”.
  • Website Traffic Checker
    Where we look
    Public rankings: Tranco, Chrome UX Report, Majestic Million.
    When it comes up empty
    Smaller and newer sites aren’t ranked. We show bands, never visit counts.
  • Tech Stack Checker
    Where we look
    What the home page loads, and the services it runs on.
    When it comes up empty
    Large sites often hide what they run on, and some software leaves no trace.
  • Website Trust Checker
    Where we look
    The site’s own pages, the domain registry and public security filters.
    When it comes up empty
    Small, honest sites often leave these details out. Signals are not a judgement.
  • Website Summary
    Where we look
    How the site describes itself, and public knowledge bases such as Wikidata.
    When it comes up empty
    Some pages give no title or description to quote.
  • Domain History
    Where we look
    The domain registry, the web archive and our own earlier scans.
    When it comes up empty
    Some registries publish no dates, and the archive misses many sites.

What the score means

It adds up what can be checked about a site: its age, what it says about itself and how it’s set up. It isn’t a safety verdict. A new, honest site starts low, and AI traces carry no weight.

The formula, point by point
PartWhat earns pointsMax
Transparencycontact page 8, contact details 4, privacy policy 6, terms 4, refund policy 4, about page 430
Track record5+ years 25, 2+ years 20, 1+ years 14, 6+ months 9, 2+ months 425
Popularitytop 10,000 20, top 100,000 16, top 1,000,000 1020
Security and setupHTTPS 8, sends visitors to HTTPS 3, always uses HTTPS 2, protective settings 215
Consistencyname matches the address 5, describes itself 2, tells search engines what it is 310

Bands: strong trust signals from 70, some trust signals from 50, few trust signals yet from 0. Anything we couldn’t check is left out, never counted against a site. Formula v1.1; every change.

How we test ourselves

We run the AI detector on 2,312 sites whose makers we know: some built with AI, some made before AI builders existed. It rarely flags a site wrongly, and it misses some AI-built ones, most often those moved off their builder.

We test again whenever we change what we look for, and publish every run, misses included.

Read the full test

What we don’t do

  • Judge the people or the business behind a site.
  • Read pages behind a sign-in, or more than the home page and the policy pages it links.
  • Guess how many people visit a site.
  • Check adult websites.
  • Keep your IP address.