The moderation pipeline
Every report goes through the same five stages before it's ever visible to the public — no listing is published automatically, and no listing skips review.
A report is recorded as "submitted" — not yet visible to anyone but our moderators.
We check whether the reported entity already matches an existing case (by name and identifying details). A match links to that case; no match creates a new pending one — either way, nothing goes public yet.
A moderator examines the report and any attached evidence, cross-references the domain/wallet/contact details against what's already known, and assigns a confidence tier before deciding whether to publish.
The reported entity can submit a response at any point — before or after publication — which is published alongside the case where received, not suppressed.
Only now does the case appear in the public database and search tools — with its confidence tier clearly shown.
What moderators actually check
A confidence tier isn't assigned on the reporter's word alone. Depending on what's available for a given case, moderators weigh:
- Multiple independent reports — do separate reporters, unconnected to each other, describe the same entity and pattern?
- On-chain evidence — wallet addresses, transaction flows, and clustering that ties funds to a known scam infrastructure
- Domain and infrastructure signals — registration history, hosting patterns, and reuse of infrastructure seen on other known scam sites
- Cross-referencing external sources — authorized scam-intelligence feeds and open-source reporting where available
- Documentary evidence — screenshots, correspondence, contracts, and bank/transaction records attached to the report
- The entity's own response — a credible right-of-reply can shift a case's status; a non-response to serious, well-evidenced claims does not automatically confirm guilt, but is noted
- AI-assisted analysis — used to help spot patterns, profile entities, and correlate evidence across cases faster; every output is reviewed and confirmed by a human moderator before it factors into a decision, never applied automatically
The three confidence tiers
The tiers form a single ladder, not three independent labels — each higher tier already includes and exceeds the one below it. A case doesn't carry two tiers at once; "Verified Pattern" already means it cleared the Corroborated bar and then some.
Reported, but not yet reviewed or independently confirmed by a moderator. This is the default status for every new case — most listings start here, and stay here until a moderator actively reviews them. It never changes automatically, no matter how many reports get linked to it.
Multiple independent sources agree, but nobody's done deeper investigative work yet. Assigned when a moderator confirms:
- Two or more unconnected reporters independently describe the same entity and tactic, with no sign they know each other
- Or a single report is backed by something concretely verifiable — e.g. a wallet address or domain that checks out
In short: several sources say the same thing, and it's credible — but nobody has personally traced, matched, or investigated it yet.
Someone actually did investigative-grade work, not just agreement-counting. Assigned when at least one of these is true:
- A completed OSINTrace investigation confirmed it
- A moderator personally traced a wallet on-chain and confirmed the funds flow to a known scam/off-ramp cluster
- The domain, hosting, or wallet infrastructure matches infrastructure already confirmed on other Verified Pattern cases
- The case matches a scam tactic so specific and well-documented that recognizing it is itself confirmation
Report count alone — even a high one — never on its own justifies Verified Pattern. A case with 10 reports and no investigative work stays Corroborated; a case with 1 report backed by a completed investigation is Verified Pattern.
What can change a listing after publication
- A credible right-of-reply response can move a case to "disputed" pending further review, or result in a correction
- New evidence — from further reports or an entity's own response — can raise or lower a confidence tier
- A listing can be corrected or removed entirely if it's shown to be inaccurate
Editorial independence
Confidence tiers and publication decisions are never for sale. A platform cannot pay to have a listing removed, downgraded, or kept off the database — the only way a listing changes is through the right-of-reply and evidence-review process above, open to anyone.