Most of this glossary explains what the industry does. This entry explains something we do, and why we accept the cost of it.
The rule
There are two scores and each has its own version of the rule, because they can fail in different ways.
The TrustAudit Index is not published until at least three of its six evidence channels have actually been examined — licensing checked against the regulator's register, the terms read for unfair clauses, the monthly withdrawal cap, the payout time, the KYC turnaround, and the wider reputational record searched across published complaint databases and review platforms. Below that the score is null: not zero, not five, not a placeholder.
The sixth channel is there because the first five are not enough on their own. A licence and a set of terms are documents an operator writes about itself, and a company can hold a genuine UK Gambling Commission licence and publish unobjectionable terms while its players spend a fortnight being asked for a fifth proof of address. That is the one thing reading its own paperwork can never show you, so we read what other platforms record as well — and count a pattern only where at least two independent ones report the same thing.
Three is the smallest number that means anything. A licence and a payout time are the two facts an operator publishes about itself on its own homepage, and taking both on trust is not an audit. The third check is the first one that requires somebody here to have opened a document, and it is the first point at which two operators can honestly come out differently.
The JuryScore is not published until there is testimony at all. The community score shrinks a small sample toward a neutral 5.5 so that two glowing or two vengeful reviews cannot swing an operator. With zero reviews there is nothing to shrink and the formula returns the 5.5 unchanged — a number that describes the prior rather than the operator. So it returns nothing instead.
Where a score is published on an incomplete file it is labelled provisional, and the outstanding checks are named on the review page. A provisional number may rise or fall as research gaps and unresolved findings are reviewed.
Why this is worth doing
The industry habit is to renormalise: score the things that are measurable on day one — site design, game count, payment methods, page speed, and present the result as a rating. It is defensible arithmetic and it produces a number, which is what everybody wants.
The number always lands somewhere around 6 out of 10. And a 6 reads to a person as "they evaluated this and it is mediocre."
The truth is different, and worse: nobody has ever tried to withdraw money from this operator, and we have no idea whether it pays.
That is the exact failure this site exists to prevent. A player deposits at a "6.3", cannot withdraw, and discovers we never had any data at all. A renormalised cold-start score is not a cautious estimate. It is a claim we cannot support, dressed as one we can.
What we show instead
The parts we do know — licence status, terms, withdrawal limits — clearly labelled as partial. And we emit no aggregate rating in our structured data for an unrated operator. We will not assert to a search engine a rating we refuse to show you.
The decision, and its costs, are written up in full at /methodology.