Where our information comes from, how a conclusion is reached, what our method structurally cannot prove, and what happens when we get something wrong.
We publish the categories of source we rely on. Individual reports footnote the specific source and its retrieval date.
| Source category | What it establishes | Grade |
|---|---|---|
| National enterprise registry | Legal existence, unified social credit code, registered capital, business scope, licence status | A |
| Judicial & enforcement records | Judgments, enforcement, dishonest-debtor listings, freeze and pledge records | A |
| Regulatory penalty registers | Market-regulation, customs, tax, environmental, fire-safety penalties | A |
| Certification issuing bodies | Whether a certificate exists, its scope, validity and accreditation status | A |
| Customs & trade data | Export frequency, volume bands, destination markets, port patterns | B |
| Social-insurance & tax filings | Proxy for real headcount and operating continuity | B |
| Supplier-provided documents | Capacity claims, equipment lists, internal quality records | C |
| Buyer references | Performance history, where the reference agrees to be contacted | B |
| Adverse media (CN & EN) | Reputational signals requiring corroboration before use | C |
| Own site observation | Physical capacity, conditions, documentary practice — audit only | A |
Every piece of evidence is graded. The grade is printed in the report so you can see how much weight a conclusion carries.
Official registry, court, regulator or issuing-body record retrieved by us or by a licensed data provider. Stands alone.
Reliable commercial or administrative data that is indirect, banded or delayed. Requires corroboration for a material conclusion.
Supplied by the subject or derived from unverified media. Never the sole basis for a positive finding — only for a question we then pursue.
A material positive conclusion — this factory can genuinely produce at volume — requires at least two independent sources, one of which must be Grade A or our own observation. A single Grade C assertion, however plausible, is reported as a claim, not a finding.
A material adverse conclusion requires one Grade A source, or two consistent Grade B sources. Where only Grade C evidence exists, we record it as an unresolved question and say so.
We do not treat silence as a clean bill of health. If a register has no record, the report says "no record found", not "no issues found".
Each of the nine modules is concluded independently before any aggregate score is calculated.
Module findings are converted to a 0–100 BV Score using published weights. Weights are versioned and change only with a dated methodology release.
Certain findings end the review regardless of total score. See the admission standards.
The score is a measurement; the recommendation is a judgement, and we state which parts of it are judgement rather than measurement.
A reviewer who did not produce the file checks sources, grades and arithmetic before issue.
These are structural limits, not shortcomings of execution. They are printed on page one of every report.
A verification reduces uncertainty. It does not eliminate it. Treat it as one input to a decision, alongside your own commercial judgement.
Desk verification relies on public and commercially available records. Supplier consent is not required to begin.
Before issue, adverse findings are sent to the supplier with the evidence supporting them and a stated response window.
The supplier's response is included verbatim. We may add a note explaining why we did or did not change the finding, but we do not edit their words.
If a supplier does not respond, the report records that the finding was notified and no response was received.
A confirmed material factual error is corrected and the report reissued. Corrections are permanent and visible on the record.
Every correction is logged and the total published in the Annual Transparency Report.
Where a methodology error affects multiple files, we withdraw and re-run at our own cost, and notify every affected buyer.
Disagreement with a judgement, an unfavourable conclusion, or a supplier's dissatisfaction are not errors.
The methodology carries a version number shown on every report. Changes are dated, described, and — where they would alter scores — applied prospectively with prior notice. We do not retro-score historical reports against a new model without saying that we have.
Published annually, whether or not the numbers flatter us: