How we test

This is the full testing protocol behind every figure on the site. Every payout time, evidence grade, and score is produced by the steps below. If a row on a money page or operator review contradicts something here, the row is wrong and we want to know — see Corrections at the bottom.

The site does not cover every crypto casino on the open web. We do not want to. The selection process is deliberate and disclosed:

1. Search-demand floor. An operator must produce enough first-party search interest (its own brand queries plus reviews-related queries) to merit a dedicated review. We do not publish stub pages for operators no one searches for.
2. License or transparent unlicensed status. The operator either holds a current licence we can verify in a regulator registry, or openly states it operates without licence and we can confirm that. Pages without verifiable regulatory status are not published.
3. On-chain footprint. For a Level A test to be possible, the operator must process crypto withdrawals on a network we can observe. Operators that exclusively offer fiat payouts are out of scope.
4. No exclusions for unfavourable results. Once an operator is selected, the result of the test is published regardless of whether it is fast, slow, KYC-heavy, or KYC-light. We do not silently drop operators that performed badly.

The site lists operators with which we have no commercial relationship and operators with which we do. The selection criteria above do not depend on the commercial relationship.

One test is one real withdrawal from a funded account to a wallet we control. The five steps are identical for every operator-network pair we publish:

1. Account setup. Fresh email, fresh password, no shared identifiers across operators. Sign-up completes only the fields the operator requires up front.
2. Deposit. Standard amount on the target network (typically 250 USDT for stablecoin tests, equivalent for BTC). We record the deposit hash and confirmation time.
3. Wager-to-clear. Where the operator requires minimum wagering before withdrawal, we play the lowest-house-edge game (typically video poker or baccarat) at flat stakes until the requirement clears. We do not chase wins.
4. Withdrawal request. Same network as the deposit, withdrawing the post-wager balance. Timestamp #1 logged at submission.
5. Tracking to wallet. Timestamps #2 (operator approval), #3 (transaction broadcast on-chain), and #4 (wallet credited) are logged. The on-chain portion is verified against the block explorer.

From the four timestamps we derive three durations. Each is reported separately on the test page — never collapsed into a single number.

Every published timing carries a grade A through E. The grade reflects how the evidence was obtained, not whether the operator is fast or slow.

Every figure on the site can be traced back to one of the following:

• Block explorers. Tronscan for TRC20, Etherscan for ERC20, Mempool.space for Bitcoin, Solscan for Solana, Blockscout for Litecoin. On-chain settlement timestamps and addresses come from these.
• Regulator registries. UK Gambling Commission Public Register, MGA License Register, Curaçao GCB, Anjouan iGaming, Isle of Man GSC, Gibraltar Licensing Authority. License numbers shown on operator pages are verified against these registries.
• Operator-controlled surfaces. Terms of service, withdrawal-limit pages, and KYC policies are fetched and quoted with the fetch timestamp. Captured text is stored verbatim; if the operator changes wording, the old capture remains on file.
• Our own withdrawal logs. For Level A tests, the operator-side timestamps (request submitted, approved) come from the operator's transaction history page screenshot captured at the moment of the action.

Anything that cannot be traced to a source in this list does not get published as a measurement. If we cannot link a claim to an explorer, registry, captured text, or own log, the claim is not on the site.

Real measurements vary. Two tests on the same operator-network pair can produce different durations because of network congestion, weekend queues, or risk-review triggers we did not anticipate. We do not average the difference away. The rules are explicit:

1. Every measurement is published in full. The test page lists the test date, amount, network, evidence level, four timestamps, and the redacted transaction hash. Older tests are not deleted when a new one is added.
2. The summary number on operator review pages is the median of the latest three Level A tests for that operator-network pair. Median, not mean, because a single 48-hour outlier should not pull the headline figure into a misleading range.
3. When only one test exists, the page shows that single number and is explicit: “based on one Level A test”. We do not extrapolate a range from a single observation.
4. If a new test diverges by more than 2× from the prior median, we re-test before updating the headline. The divergent test stays visible in the test list with a note explaining the investigation status.
5. If conditions changed — operator changed payment processor, network had a documented congestion event, account was flagged for review — the divergent test is flagged but not hidden, and the change in conditions is recorded as part of the test record.

The PayoutDB Score visible on operator review pages and the best-of leaderboard is a weighted sum of five components, on a 0–100 scale. The weights are fixed and identical for every operator. We do not hand-tune individual scores.

Best-of pages rank operators by the score above, then by tier of evidence (A first, then B, C, D, E), then alphabetically. We do not award stars, badges, or “editor's picks.” An operator without a Level A test cannot rank above one with a Level A test at the same speed — the evidence-quality component prevents it.

For every Level A test, the full transaction hash is stored in our database but never appears in public HTML, CSV, JSON-LD, or the sitemap. What appears on the test page is the redacted form — first 4 and last 4 characters — and a server-side redirect at /verify/<test_id>/ that 302s to the block explorer. The redirect route is disallowed in robots.txt so the hash is not indexed.

Top-priority operators are re-tested quarterly. Other tracked operators are re-tested annually or when a regulator action, ownership change, or pattern of user reports warrants. A test older than 12 months is still visible but its weight in the score is tapered linearly toward the next-lower evidence tier. The historical record stays on the entity page changelog and is not overwritten silently.

A single payout test reflects one wallet, one amount, one moment in time. It does not prove that the operator will always behave this way. We avoid absolute language and reject submissions that rely on it. The evidence-based phrasing on the site is deliberate: it reports what we measured under the stated conditions, nothing more.

The site earns commission from some operator links — every such link is marked rel="sponsored" and disclosed on the page that contains it. The presence or absence of an affiliate relationship has zero input into the score, the evidence grade, or the ranking position. We have published Level A tests for operators we have no commercial relationship with, and Level D / no-test status for operators that pay us. The score weights above do not include “commercial fit” as an input.

This site is informational. It does not solicit play from any specific jurisdiction. Every operator page carries the 18+ notice, a link to BeGambleAware, and — where the offer is region-restricted — the geographic applicability. We do not list operators we know to be operating without licence in a regulated market we cover. If you are reading this in a country where online gambling is restricted, we do not encourage you to play.

If a published figure is wrong, write to [email protected] with the test ID and the specific claim you are contesting. We respond within 10 business days. Operators contesting a test must supply timestamps and a transaction hash for an alternative withdrawal under documented conditions — claims without evidence do not move grades. Any change to a published figure is logged in the changelog on the affected entity page.