How we rate casinos
Every verdict on this site is assembled from the same four evidence classes, in the same order of authority. When a class is missing, the page says TODO in its place — the review template literally refuses to fill those sections any other way.
The four evidence classes
| Class | Source | Example |
|---|---|---|
| 1. Regulatory record | The licensing authority’s register (MGA or otherwise), licence conditions | Licence of record, legal entity, ADR body |
| 2. Our test log | Timestamped deposits, withdrawals, KYC runs on funded accounts | Measured payout hours in the Payout Speed Index |
| 3. Primary documents | Operator T&Cs, provider game sheets | Bonus wagering terms, default RTP values |
| 4. Verified reader reports | Screenshot-evidenced submissions, re-verified | Reduced-RTP sightings, payout timings |
Operator marketing pages are not an evidence class. They’re the thing the classes above exist to check.
What a review must contain before it publishes
- Licence of record — authority and licence id, linked to the authority’s register where one exists (hard gate — no verified licence, no page). You can repeat the check yourself in a minute with the licence verification guide.
- A completed hands-on test log: real deposit, real withdrawal, timed verification.
- Bonus material terms adjacent to the offer — wagering, max bet, eligible games, expiry, max cashout — never behind a “T&Cs apply” link.
- Payment acceptance verified per rail from a Maltese account, not copied from a cashier FAQ.
- A named author and a “last verified” date.
Reviews missing any of these render with visible TODO flags and stay out of the search index until complete.
Verification cadence
- Payout measurements: rolling retest cycle per rail ; stale data ages out of the index rather than lingering.
- Licence checks: re-verified at every content update touching the operator .
- Bonus terms: re-read on every offer change we detect; the “last verified” date on each page is the honest freshness signal.
The commercial firewall
We earn affiliate commissions (full disclosure). The firewall is structural: measured data enters pages from the data files the test team maintains; commercial relationships have no write access to those files, no influence on the index methodology, and no veto on a negative verdict. If an operator’s numbers are bad, the numbers publish.