Publication standard / 03

Editorial standards

These rules govern what can be claimed, how commercial relationships are separated from evidence, and how a dossier changes when stronger information arrives.

AccuracySpecific evidence outranks confident language.
IndependenceCompensation cannot buy score or verdict.
RevisionMaterial corrections remain visible.
01

Claim discipline

Every material statement must have a defined basis.

Numerical limits, operator identity, licence status, KYC triggers and withdrawal timing must be traceable to a named source or explicitly labelled as an observation. We distinguish operator claims from regulator-confirmed facts and independent reports.

Headlines and short answers must reflect the evidence boundary. We do not translate an unpublished cap into unlimited capacity, routine processing time into a guaranteed review ceiling, or a successful transaction into proof of future solvency.

02

Independence

Editorial and commercial decisions use separate inputs.

Affiliate relationships can influence which alternative receives a commercial placement. They cannot add score points, improve confidence, remove negative terms, change source state or reorder a documentary ranking.

Commercial links are marked with sponsored and nofollow relationship attributes. Readers can inspect the affiliate disclosure for the funding model and the methodology for ranking inputs.

03

Fair process

Operators may supply evidence and a right of reply.

An operator may challenge a factual statement, identify a newer document or explain an account-specific rule. We request a public source or a reproducible, dated record and assess it under the same hierarchy as existing evidence.

A response is not accepted merely because it is authoritative in tone. Material claims require enough detail to identify the account state, asset, network, amount, period, jurisdiction and effective date.

04

Tools and AI

Automation assists collection; it does not erase editorial accountability.

Automated tools can test URLs, detect changes, normalize fields and identify contradictions. AI-assisted drafting can help organize research. Neither is treated as a primary source, and generated document maps are not counted as independent screenshots.

A human-readable evidence trail must remain available. Transport failures, content mismatches and uncertain extraction are exposed for review rather than converted automatically into factual conclusions.

05

Risk language

Verdicts describe evidence and suitability, not criminal guilt.

Terms such as unsuitable or insufficient evidence refer to the selected high-limit scenario. They do not allege fraud. Scam, theft or insolvency claims require a substantially higher evidentiary threshold than an opaque withdrawal policy.

We avoid guaranteed-safe language. A tested commercial alternative can still change terms, restrict a jurisdiction or require compliance checks.

06

Release gate

A page must pass editorial and technical checks.

Before publication, a dossier is checked for a unique H1 and title, a clear short answer, source-linked material numbers, non-duplicated copy, descriptive image alternatives, responsive tables, working breadcrumbs and valid structured data.

Updates that alter a numerical conclusion trigger a scenario, score, confidence, metadata and internal-link review—not a silent sentence replacement.