Editorial Policy

Assertions on this site are made in our voice, because we have done the work to stand behind them. Sources appear as evidence, not as shelter: when we write that a provider holds a license, we checked the register, and the citation is there so you can too. Where we could not verify a claim, we say so explicitly, and phrases like self-reported, provider-stated or not register-verifiedare our verdicts on the evidence, not decoration. Attribution of the "according to X" kind is reserved for two cases: claims we could not independently confirm, where the attribution is a warning, and figures that are inherently non-public, where it is the honest limit of anyone's knowledge.

No review is sponsored and nothing is ranked for money: the scoring is built from data under a published methodology, providers cannot pay to enter the catalog or improve a score, and the details live in our methodology and advertising policy. We publish without bylines, deliberately: reviews are produced by an editorial process rather than a personality, every substantive claim is checkable against the cited source, and the site stands behind the work as one voice.

Load-bearing facts, the ones a reader might act on, must trace to a primary source: a regulator's register, a legislature's record, a network's published rules, a company's own filing. Trade press can point us at a story, but it does not get to be the evidence, because trade-press dates and numbers drift, and our corrections log below documents cases where they drifted onto our pages before we tightened this rule.

License claims get the strictest treatment. We audit them against the registers themselves, entity by entity, because registers are brand-blind and marketing is not: the series so far covers the ESMA CASP register, the UK FCA register, the Brazilian central bank and six US state gaming registers. When a register cannot be publicly checked, we say "not checked", never "not found", and claims that fail verification are reframed on the provider's review the day the audit closes, in both directions: overclaims get flagged, and providers whose cards undersold their real licensing get upgraded with the same diligence.

Numbers we cannot trace do not run. That includes statistics whose trail ends at content farms, single-vendor claims with no second source, and round numbers that every outlet repeats and none can origin. Where our own catalog is the dataset, figures are computed from it at build time rather than typed in, so they cannot drift from the underlying data.

The visible "Updated" date on any page is the date its content substantively changed, and it is the same date the sitemap and structured data carry. We do not scatter fresh-looking dates across unchanged pages. A date only moves when the content does. Provider reviews run on a quarterly re-verification baseline, 77 providers deep, with an event-driven layer on top: a register audit, a rule change or a market launch triggers same-day updates to every affected page instead of waiting for the next scheduled pass.

When we find a factual error in something we published, we correct the text, move the page's modification date, and record the error in the public log below: what we said, and what is correct with the source that settles it. Most entries come from our own verification work, which is the point of doing it. Errors of judgment (a score you disagree with, an editorial call) are not corrections. Errors of fact are, and the difference is whether a primary source settles it.

If a provider disputes a factual claim on its review, we investigate against primary sources. If they are right, we correct it and log it here. If the dispute is with our scoring or framing, we explain our reasoning and the score stands. Spotted an error, on any page? Tell us, ideally with the primary source, and it gets the same treatment.