A Floath Score is not a summary judgement written first and justified afterwards. Each operator is scored 0–100 in seven categories, each category carries a fixed weight, and the overall figure is the weighted mean of those seven numbers. Nothing else feeds it — not traffic, not commission, not how long a brand has been on the site.
The overall score is derived, never stored. There is no field on a casino record that holds it. Every list, badge and comparison row you see recalculates it from the categories and the weights that are live at that moment, which is what makes the rest of this page checkable rather than decorative.
The seven categories and what each one weighs
| Category | Weight | Evidence a score has to rest on |
|---|---|---|
| Safety and licensing | 25% | The regulator's own published register; the company statement in the operator's footer; the limit, time-out and self-exclusion controls actually present in a logged-in account. |
| Withdrawals | 20% | Stated operator processing time recorded separately from banking time; published limits and fee schedule; the point at which identity documents are requested. |
| Bonuses and value | 15% | The full terms page: multiplier, whether it applies to bonus or deposit-and-bonus, max bet, game weighting, excluded games and expiry — as fields, not as quoted offer copy. |
| Game selection | 15% | The lobby itself: studio list, category counts, whether RTP is discoverable per title, and whether search and filters work on a phone. |
| User experience | 10% | Registration completed end to end, then the cashier and the limit screens on a phone. Dark patterns around bonus opt-in or limit removal are marked here. |
| Payments | 10% | The cashier's method list per accepted market, its stated fees, minimum deposit, and whether withdrawals return on the rail used to deposit. |
| Customer support | 5% | Live chat and email contacted with a specific question about a term. The wait and the answer are both recorded; a pasted policy scores worse than a cited clause. |
Safety carries the heaviest weight because it is the only category where a bad result costs you the whole balance rather than a few hours. Support carries the lightest because good support is a repair for problems the other six categories should have prevented.
The arithmetic, in one line
Multiply each category score by its weight, add those up, divide by the total of the weights used, and round to one decimal place. Badges show the whole number; the decimal appears only in the breakdown table on a review. A category with no score and a category with a weight of zero are both skipped, and the divisor shrinks accordingly — so a partially scored record is never quietly penalised for the gaps.
| Band | Range | What it means |
|---|---|---|
| Excellent | 90–100 | Strong in the heavy categories, with no serious weakness anywhere. |
| Strong | 80–89 | A good operator with one or two real trade-offs stated in the review. |
| Fair | 70–79 | Usable, but something specific is wrong — usually payout speed or bonus terms. |
| Below par | Under 70 | At least one category is bad enough to change the recommendation. |
Why the weights are data, not code
The weights live in the content model and are editable from the admin panel. That is deliberate, and it has a consequence worth stating plainly: change one weight and every ranking on Floath changes with it. The default sort on every directory and comparison page is score, so the order of the operator list, the badge on each card, the contribution column in each breakdown and the composition of the curated pages all move in the same edit.
Take two operators and look at only safety and withdrawals in isolation. Operator A scores 60 on safety and 95 on withdrawals — fast money out, thin protection. Operator B scores 95 on safety and 70 on withdrawals — well regulated, slower to pay.
| Weighting | Operator A | Operator B | Result |
|---|---|---|---|
| Safety 25, withdrawals 20 (current) | 75.6 | 83.9 | B ahead by 8.3 |
| Safety 10, withdrawals 35 | 87.2 | 75.6 | A ahead by 11.6 |
That swing is the honest reason the weights are published rather than described. They encode a view — that being paid reliably matters more than being paid quickly — and you are entitled to disagree with it and read the category scores directly instead. Every review shows all seven, with the points each one contributes.
Confidence, and what it is telling you
- High — every scored field has a recorded source and was checked in the last review cycle.
- Medium — most fields are sourced; a small number are pending re-verification.
- Low — several fields still need verification, and the score should be read as provisional.
Confidence is separate from the score and never adjusts it. A well-run operator that has not been re-checked recently keeps its number and loses confidence, because quietly shading a score to express doubt would hide the doubt inside the thing you are trying to read.
How often a record is re-reviewed
Every record carries four dates: first published, last edited, last reviewed and next review due. The next-review date is set between 60 and 150 days after the last review, depending on how quickly the fields move — cashier terms and bonus terms are re-checked on the shorter cycle, game libraries and regulator descriptions on the longer one.
| Status | Trigger | What it means for the record |
|---|---|---|
| Current | More than 30 days until the review date | Nothing outstanding. |
| Review soon | 30 days or fewer until the review date | Queued for re-check; the record is still trusted. |
| Outdated | The review date has passed | Flagged on the admin freshness dashboard until someone re-checks it. |
| Verification required | The record's provenance is set to verification-required | Overrides the dates entirely. Nothing on it should be read as confirmed. |
A record does not fall out of the site when it goes outdated — it stays visible with its date on it, because hiding a stale review is worse than showing one that admits its age. The dashboard exists so the queue is generated rather than remembered.
How to challenge a rating
Anyone can challenge a score, including the operator being scored. The route is the same either way, and it goes to editorial@floath.com.
- Name the operator, the category and the specific figure you say is wrong — for example the withdrawal processing time, or whether wagering applies to deposit plus bonus.
- Send the evidence: a link to the operator's own terms or cashier page, or the entry on the regulator's public register. A screenshot without a URL is not enough to change a field, because nobody can re-check it later.
- The reviewer named on the record re-checks the field and writes a source row recording what was read and when.
- If the field changes, the category score is recalculated, the overall score moves with it automatically, and the record is re-stamped with new edited and reviewed dates.
- If the field does not change, you get the reason and the source that was used, so the disagreement is at least about the same document.
We aim to acknowledge a challenge within five working days and to resolve one that turns on a single checkable figure within ten. Challenges that would change a number are prioritised over everything else in the queue. Two things are not on the table: a score cannot be changed because an operator asks for it without evidence, and no operator gets to edit or approve review copy before it is published.