Every fact sourced & dated — independent, evidence-first comparisons MethodologyGlossary
Home/Methodology
ProxyMojo

Methodology

This directory is evidence-first. Every material fact about a provider — a price, a country, a feature, a test result — must resolve to a source, a date, an evidence label and a named review decision before it appears publicly. Where we do not have reliable information, we say Unknown. We never convert a missing value into an estimate.

Evidence labels

  • Provider stated — a named official source (pricing page, documentation, policy page) says it, with a capture date.
  • Observed in controlled test — our own authorised test measured it, against endpoints we own, under a published test specification.
  • Editorial assessment — a human reviewer's opinion, labelled as such.
  • Derived calculation — a transparent calculation from cited facts (for example, effective price per GB).
  • Unknown — reliable information is absent. It stays visible as Unknown.
  • Superseded — an earlier approved fact kept in history with its validity dates.

How data enters the directory

Research drafts — whether entered manually, imported from CSV, or collected from approved official provider pages — never publish automatically. Each candidate fact is normalised, compared field-by-field against the current approved record, and queued for human review. A reviewer other than the submitter must approve every price, coverage claim, company fact and benchmark result before it goes live. All approvals, rejections and rollbacks are audit-logged.

Location claims

We record four different things separately and label which one we mean: exit-IP availability, targeting granularity (country, region, city, ASN, carrier), physical server location, and observed country match from our own tests. A provider's global pool size or country count is never presented as country-specific supply.

Benchmarks

Controlled tests run only through provider accounts whose terms permit testing, against rate-limited endpoints we own. Published results always show the test spec version, window, sample size, runner region and limitations, and we do not publish scores from fewer than 100 comparable attempts. We do not test or score bypassing of third-party platforms, CAPTCHAs, streaming services or security controls.

Affiliate disclosure

Some outbound provider links may earn this site a commission. Commissions never influence rankings, benchmark results, comparison outcomes or editorial conclusions, and affiliate relationships are disclosed near commercial links.