Where the data comes from, and what is refused
Direct answer
A directory is a set of claims about third parties. This page states where each claim comes from, and which claims we decline to make at all.
Key facts
- A directory is a set of claims about third parties, and a claim is only worth what its source is worth.
- What is refused, and each refusal is a decision rather than an omission.
- Unverified records are absent, not pending.
- No invented numbers. Where a value is unknown, the field says so.
- No ratings. No star scores and no fabricated reviews.
A directory is a set of claims about third parties, and a claim is only worth what its source is worth. This page states where each part of the data comes from, in the order the checks run, and which claims we decline to make at all.
Where each part of the data comes from
| Item | Source | What it can and cannot establish |
|---|---|---|
| Legal name, credit code, operating status, insured headcount | Public registration lookup | That the entity exists and is active. Not that it makes what it claims |
| What the factory actually makes | Registered scope, own product pages, catalogue, certifications | That tackle is the main business. Not that capacity is what is stated |
| Whether it exports, and to where | Customs export records | That goods shipped under that entity. Not how well they were made |
| Capacity, MOQ, lead time, price basis | Structured interview with the factory | What the factory states. Not a measurement we took |
| Reference price bands | Listings sampled on this site | A plausible range for the product. Not a quotation |
| Cluster context | Public industry reporting and trade statistics | Scale and shape of a supply base. Not the quality of one factory |
| Dates, re-check status, machine-readable files | This build | When each record was last checked, and in what form it was published |
The pattern is deliberate. The first three sources are outside our control and can be re-read by anyone; the fourth is a claim by an interested party, and every field drawn from it is written as what the factory states rather than as a finding. A buyer reading a capacity figure needs to know which of those two things it is.
What we publish, and what we refuse
What is published: the verified factory records with their review dates, the sampled product listings with reference bands and MOQ ranges, the four verification gates as a published standard, the reference sources behind every claim, and the same content as machine-readable files at stable URLs.
What is refused, and each refusal is a decision rather than an omission.
Unverified records are absent, not pending. Listing a factory as under review tells a buyer it is worth a look, and that is a recommendation we have not earned the right to make. Records that fail a gate are recorded internally with the reason, and the count of withheld records is published on the directory page so the gap is counted rather than hidden.
No invented numbers. Where a value is unknown, the field says so. This costs us something real: a record with two empty fields looks weaker than a record with two plausible estimates beside it. But an estimate is indistinguishable from a measurement once it is on a page, and the entire value of a verification record is that the difference is visible.
No ratings. No star scores and no fabricated reviews. A score hides which dimension was weak, which is the opposite of what a verification record is for — the gates state what was checked and what failed, and a reader can disagree with a specific finding rather than with a number out of five.
No paid placement. No sponsored listings and no ranking for sale. A directory that orders its results by who paid is a directory whose ordering carries no information, and the ordering here is the reading order of the method.
No personal data. Business contact information only, in line with applicable data protection rules.
The four checks behind every published factory
Registration, main business, export records, and a direct interview — run in that order, published as a standard, and enforced by the build rather than by an editor. A record that fails any one of them does not appear. The detail of what each gate checks, what disqualifies a record and what each gate costs is published on its own page.
Two consequences of running it this way are worth stating. The directory is smaller than the platforms list, because an interview does not scale the way a form does. And a factory that fails is absent rather than downgraded, which means the directory cannot be sorted into "good" and "less good" — everything in it passed the same four checks, and the differences between records are in the fields rather than in a grade.
What would make us wrong
Three conditions, stated so that a reader can judge the site against them.
If a record is published and the factory turns out not to exist as described, gate 1 failed and the standard is wrong. If a published factory systematically understates lead time or overstates capacity, gate 4 is asking the wrong questions — it currently asks for numbers, which is the strongest available defence, but a stated number is still a statement. And if the reference bands drift far from what buyers actually pay, the sampling behind them is unrepresentative rather than the method being unsound.
Each of those is checkable, and a buyer who finds one should say so: the contact route is published, and a record that fails a re-check is withdrawn rather than quietly aged out.
What do we refuse to publish?
- A factory that fails any one of the 4 verification gates.
- A record marked "pending" —a failed record is absent, not queued.
- A price without its unit and its basis.
- A certification the factory cannot evidence.
- A supplier rating or a review score —we do not collect them, so we do not display them.
Where this sits in the process
This page is one topic out of the verification standard. The overview is About us, and the other topics on this page are listed there.
Sources
- This page restates our own published scope and compliance decisions; no external source is cited