The ranking is the product, and the product has a price list
WATT Global Media’s rate card for the Petfood Industry Directory of Suppliers states what the money buys. A basic listing is free and covers up to ten category listings. An enhanced listing costs USD 1,000 in print, USD 1,500 online, or USD 2,500 for both, and its first listed benefit is not reach, a logo or a video. It is position: the company name “appears at the top of each product category in alpha order” 1.
That sentence is the most useful piece of evidence in this corner of the market, because it settles a question buyers ask constantly and answer wrongly. When a sourcing manager opens a category page in a trade directory and reads the first six companies, they are reading a price ladder with an alphabetical sort applied inside each rung. Payment determined who is at the top. Alphabet determined the order among those who paid.
Thomas does the same thing in a different vocabulary and admits it just as plainly. A case study on Thomas’s own blog describes an advertiser listed in more than 2,500 Thomasnet categories, and says that “by opting for the Premium Vertical Sponsorship package, Kershaw ensures that Zooby will show up at the top of those search results” 2. Two unrelated directories, on two continents, in two industries, sell the same slot.
The finding is not that directories are corrupt. Paid placement is disclosed, sometimes prominently, and a buyer who reads the media kit can see it. The finding is that a ranking is not an editorial judgement about quality. It is a commercial product with a published price, which some suppliers buy and others write off as unfair.
Once the order is sold, the directory’s remaining job separates two objects buyers treat as one. A platform lists suppliers. A database records them. A listing is a claim a supplier makes about itself, formatted to a template and often paid for. A record is an entry created by somebody else while doing something else: a registrar recording a company, a customs service recording a consignment, a competent authority recording an approved plant.
That distinction, not the size of the platform, determines evidentiary value, and it explains the migration. Buyers do not leave general marketplaces because the photographs are bad. They leave at the moment their own liability becomes specific.
Four families of record, and what each one holds
The vertical supply of supplier information sorts into four families. They differ by who wrote the record, and that determines how much weight it carries.
| Family | Example in this article | Who wrote the record | What it establishes | What it cannot establish |
|---|---|---|---|---|
| General B2B marketplace | Alibaba, Made-in-China, Global Sources | The supplier, plus an inspector the platform hired | A company exists, pays for a storefront, and in some tiers passed a document or site check | Capacity, product conformity, continuity |
| Vertical or industry directory | WATT’s Buyer’s Guide, Pet Food Processing Buyers’ Guide | The supplier, ranked by a sales process | A company claims a category and, usually, advertises in it | Anything operational |
| Company registry | Kompass, Europages, OpenCorporates | A database operator, from registers and self-submission | Legal identity, address, declared activity, sometimes filings | Capability, quality, solvency |
| Primary record | EU establishment lists, customs manifests, GS1 | A regulator, a customs service, a standards body | One specific fact, at one date, under a legal obligation | Everything outside that fact |
The families are graded by who bears the cost of being wrong. A supplier who overstates capacity on a storefront risks a bad review. A customs service that mis-records a consignee risks a duty assessment. A competent authority that lists a plant which cannot meet the rules risks a border refusal traced back to its own inspection. Reliability tracks the recorder’s exposure, not the recorder’s prestige.
They are also not substitutes, and the migration between them is not a queue from worse to better. A customs record showing a factory shipping to a named brand still does not say whether that factory can hold a new specification. An EU establishment number still does not say who owns the company. Each record closes one question and leaves the rest open.
The general marketplace: the badge is a tier, and the tiers differ
Most buyers start in a horizontal marketplace, and the honest description of what those platforms offer is more sophisticated than the critique allows. They have built verification products. The problem is not that verification is absent. It is that the word covers at least three depths of check at three prices, and the badge rarely says which one the buyer is looking at.
Alibaba publishes its own gap analysis
Alibaba’s own explanation of its tiers is the clearest document in this family, because it prints the comparison as a table rather than a claim. It separates a Gold Supplier tier, described as “A&V and on-site Check”, from a Verified Supplier tier, described as “In-depth inspection by independent third part institution” 3.
The table underneath is the interesting part. Against a long list of attributes, the Gold tier carries a tick only for business licence and ownership status, contact information, and company or factory location. Business type and company structure, inspection videos, export markets, main clients, R&D staff, production capacity and flow, production machinery, subcontractors, quality management and QC processes, product certification, testing capabilities and after-sales service all read with a dash 3. The platform is telling the buyer, in its own layout, that a paid storefront at the lower tier verifies a legal identity and an address, and nothing else.
At the top tier a third-party inspection covers those same fields. What it does not do is convert the visit into a durable record: an inspection describes the factory as it was that week, so a buyer placing a repeat order eighteen months later is reading a historical statement with no visible date.
Made-in-China has run the same product since 2007
Made-in-China’s Audited Supplier programme is older than most of the platforms’ current branding. The company states that it launched the programme in 2007 and “entrusted SGS, Bureau Veritas, TÜVRheinland and CTI” to verify suppliers and produce an audit report, and that the service was “upgraded… in June 2023” 4. The process as published runs through a pre-audit meeting, a document audit, an on-site verification covering raw-material warehouse, production plant, testing laboratory and finished-goods warehouse, and a final confirmation. The report is organised around eight capability areas: general information, export trade capacity, manufacturing capacity, service capacity, quality control, R&D, financial capacity and sustainable development capacity. Reports are published free online 4.
Read as evidence, that is a substantially stronger artefact than a storefront. It is a site visit by a named international inspection firm, covering production and quality systems, with a financial element and a public report. Read as a system, it has three limits that matter more than the scope.
It is optional and seller-funded, so its presence is a marketing decision and its absence is not a finding about the factory. It is point-in-time, so it decays between audits without saying so. And it sits inside a marketplace where the same page is also selling membership, which means the buyer has to separate the audit from the advertising on a single screen.
Global Sources separates the registration check from the licence check
Global Sources has the most granular published taxonomy in the general tier, and it is worth reading in full because the distinctions are the ones buyers need.
A Verified Supplier, in the company’s definition, is a supplier or exhibitor “whose business registration details have been verified by independent third parties”. A Verified Manufacturer is a supplier “whose business scope of operations is government-licensed to manufacture goods”, with credentials verified by third-party agencies. An O2O Manufacturer is a verified supplier that also exhibits at Global Sources trade shows, and that status involves “face-to-face verification” conducted by Global Sources staff rather than an outside agency 5.
Underneath those badges sits a Business Registration Profile, which presents “business information obtained from relevant government agencies, checked and provided by independent third parties”, and lists what it contains: year established, authorised capital, legal representative and business scope 5. That field list is the most useful thing on the page. Year established is a proxy for trading history. Authorised capital is a registered number, not paid-in capital, and the two are not the same in any jurisdiction that distinguishes them. Legal representative identifies a person who can be checked. Business scope is the registered activity list.
Verified Manufacturer is the strongest of the three, because it tests something a supplier cannot invent. A licence that includes manufacturing is a government statement that the entity is licensed to manufacture, and it is the cheapest answer to the question “is this a factory or a trader”. It is still not a capacity measure: a registered scope says what the entity may do, not what it does, at what volume, to what standard.
Global Sources then makes the crucial distinction explicit, and charges for it. The Supplier Capability Assessment is “an independent, third-party report based on an onsite visit”, covering operations, production capabilities, QC processes and management, and it is “an optional paid service for all Verified Suppliers” 5. So the platform’s own architecture tells the buyer where the free evidence stops and the paid evidence starts. A registration check is included. A site visit is an add-on.
The company also publishes the boundary of its own responsibility, in a sentence that deserves to be quoted in any buyer training deck: it rejects or pulls down suppliers proven to be fraudulent, but is “unable to intervene or make judgments in commercial disputes between buyers and suppliers” 5. That is a platform describing itself as a registry with a takedown mechanism, not as a guarantor.
One note on scope. Global Sources publishes no supplier count this exercise could verify, and its homepage returned a bot-protection interstitial rather than a page on the access date. No count is cited here, which makes the point: buyers comparing platforms on supplier numbers are comparing figures no auditor has seen.
The registries: identity, classification, and the limits of a directory
Between the marketplace and the regulator sits a layer that most buyers use daily and almost nobody audits: the commercial company registry. Its function is to establish that a company exists, where it is, and what it says it does. Its output looks authoritative because it is formatted like a reference work. Its evidence value is narrower than its appearance.
Europages: activity is declared, not tested
Europages, operated by Visable, presents a database of “approximately 2,600,000 companies mainly from european countries”, organised across 26 business sectors, 4,500 headings and more than 90,000 keywords, with an interface in 21 languages 6. A supplier profile is free to create, and the platform markets the profile to suppliers on the basis of reaching “1.2 million monthly B2B buyers” and “128,000 daily searches” 7. A premium profile is the paid upgrade.
The classification scheme is the part buyers rely on, and its structure is visible in the search interface. A buyer can filter by supplier type: Manufacturer/Producer, Customer-specific manufacturer, Distributor, Service provider, Wholesaler 8. Read as taxonomy, that is a sensible set. Read as evidence, it is a set of labels a company applies to itself, and the platform allows more than one at a time.
A search for pet food returns the illustration. One European company carries four types simultaneously, appearing as Manufacturer/Producer, Customer-specific manufacturer, Distributor and Wholesaler in the same result 8. Another listing returned for the same query is labelled Distributor and sells plant extracts, essential oils and cosmetic products alongside a pet food line 8. Neither is misconduct. Both show why “Manufacturer” in a Europages filter cannot exclude a trading house, and why a buyer who treats the label as a screen is screening on a self-description.
The same results pages show a second drift. Prices and minimum order quantities now appear in search results, in euros and in boxes, pieces or pallets 8. A registry that displays price and MOQ has taken on the mechanics of a marketplace while keeping the vocabulary of a directory, and the buyer has to decide which of the two they are reading.
The platform’s own badge closes the loop on how much a visual mark means. The Europages badge is described by the company as “piece of code that contains a link to your company profile”, which the supplier installs on its own website, and its stated purpose is to “highlight your expertise and to strengthen your credibility” 9. The badge is a link, self-installed, pointing at a profile: a reasonable marketing tool, and not a verification mark.
A separate “Verified” attribute does exist on company profiles and appears as a filter in search results 8. What it attests is not published on the pages reviewed. This article therefore records it as undefined rather than assuming either the generous or the cynical reading.
Kompass: the classification is the product
Kompass reports a larger company population than any other registry in this article, and its homepage puts three numbers side by side: “+57M Companies”, “70M Audience” and “28M Buyers looking for suppliers”, across “+70 Countries”, with 10,000 clients and 300 employees 10. Three populations on one page, none of them defined. A company count, an audience count and a buyer count are not the same measurement and cannot be substituted for one another; a buyer citing “57 million suppliers” has changed the unit twice.
What Kompass sells, beyond lead data, is a filing system: 67 activity sectors in 15 generic groups, 3,000 activity branches and 55,000 products and services, in more than 70 countries and 26 languages 11. The company argues this is more granular than CPV, SIC or NACE, which it says run to a few thousand references. For a buyer enumerating an obscure input across many countries, that granularity is the product: it is how you list extruder-die makers or gelatine-binder suppliers in eight markets before you have any other way to find them.
The analytical discipline is to use the classification for enumeration and not for qualification. A company filed under a branch has been placed there by a data process, and the placement answers “what is this company’s line of business as recorded” rather than “can this company make my product”. Kompass’s own sales material frames the database as a prospecting and visibility tool for sales teams 10, which is a fair description of what it is optimised for.
Thomasnet: region-locked, and “verified” is a subscription
Thomas is the North American counterpart, and it is the clearest case in this article of a directory whose model changed without its reputation changing with it.
The scale claims are large and old: “more than 500,000 North American suppliers”, “over 80,000 industrial categories”, “over 125+ years”, tracing to a founding in 1898 as the Thomas Register 122. The buyer-side tooling is real. A buyer can filter by location, certification and business type, compare up to five suppliers, and send one brief to up to 20 at once 13. For the upstream half of a pet supply chain (pouch converters, injection moulders, co-packers) that beats any pet-specific directory, because pet-specific directories do not list them.
Then comes the word. Thomas’s About page describes access to a “network of vetted industrial suppliers” 12. The Help Center defines what the badge behind that phrase means: Thomas has “confirmed the company operates in North America, provides accurate business information on the Thomasnet.com supplier discovery platform, has undergone a complete review and validation of all products and services currently offered, and is open to quoting new opportunities” 14. The three key features are validated contact information, validated products and services, and an activated contact option.
That is a data-quality and responsiveness check. It is a genuinely useful thing to know before spending a week on email, and it is not a factory check. No audit standard, no certification verification and no financial review appears in the definition.
The mechanism is where the argument lands. Thomas’s supplier page invites a company to claim a free profile and then to “upgrade to a Verified subscription” 15. The Verified badge is a deliverable of a paid program: the Verified Core Program “includes” the badge, profile creation, category classifications, program points for audience targeting, and reporting, with additional program points sold “in packages of 5 at $115/month” 16. The badge is not conferred by an independent body. It is a product line, delivered by a “Supplier Analyst team” the platform employs, and priced.
Nor is the badge the only thing on sale. The same blog that documents the sponsorship package also records that an advertiser can be listed in more than 2,500 categories and can buy the top of the results in them 2. Buyers who filter by certification before reading the ranking are, in effect, refusing the product being sold to the supplier. Buyers who scroll from the top are consuming it.
The regional limitation is structural rather than commercial. Thomas’s badge requires the company to operate in North America, and its database is explicitly North American 1214. It cannot find an Asian factory, and it is not trying to. Its use for a pet brand is to assemble the supply chain around the factory, in the market where the brand sells, which is where packaging, filling and logistics costs are actually set.
The pet vertical proper: two databases pointing in opposite directions
The pet industry’s own directories are small, specialised, and frequently misread, because the most authoritative one does not list the companies most buyers are looking for.
WATT’s Buyer’s Guide lists the suppliers of the manufacturers
The Petfood Industry Directory of Suppliers, now presented as the Pet Food Supplier Buyer’s Guide, is an online database of “equipment, ingredients, packaging, services and testing suppliers for the pet food industry”, serving domestic and international markets, from a publisher that has covered the sector since 1917 17. Its category list runs through analysers, extruders, dryers, coaters, palletisers, palatants, vitamins and premixes, cans, pouches, co-packing and custom manufacturing, and laboratory analysis 1.
That is an upstream directory. It lists the companies that sell to pet food manufacturers. A brand owner uses it to find a co-packer or an extruder supplier; a retailer would find nothing in it at all; and a factory looking for customers will find only its own suppliers.
The same publisher runs a second database pointing the other way: “select data sets from the Top Pet Food Companies database”, collected as part of research for a ranking of manufacturers 17. The pet vertical’s most established information business therefore keeps two directories, one of the suppliers to manufacturers and one of the manufacturers. The companies in the first sell, in aggregate, to the companies in the second, and a buyer who confuses them has the direction of trade backwards.
The rate card makes the ranking legible. Basic inclusion is free, so absence from the guide is not evidence that a supplier does not exist: the publisher has no reason to exclude anyone who fills in the form. Position costs money, and the paid benefit is stated as position at the top of the category, alphabetically ordered among the payers 1. Online enhancements run for one year from purchase with self-service updates, which dates every entry: it is as current as the last time the supplier logged in 1.
The card is dated 2020 and no current one was published on the pages reviewed, so the prices are 2020 prices. The structure of the offer is the finding, not the number.
The Pet Food Processing guide states the model in the first paragraph
The sister publication in the same segment, Pet Food Processing, publishes an annual buyers’ guide whose contents are listed without euphemism: an alphabetical suppliers directory, a products and services buyers’ guide, a state of the industry report, a new products overview, and associations and academia listings 18.
The advertiser proposition is equally direct. Advertisers get their brand in the print edition, the digital edition and the searchable online directory, with contact information, website and logo included, and the guide is described as “positioning advertisers at a critical point in the purchasing process” 18. Close date 24 August, publication September/October.
Nobody is hiding anything here. The entry standard is alphabetical; the visibility standard is advertising. A buyer using such a guide as a shortlist generator is using an advertising product as a shortlist, which is acceptable if the buyer knows it and dangerous if they do not.
One structural point applies to both titles. Their directories are annual or annually refreshed, which makes them snapshots of a marketing relationship rather than a live register. A company that stopped trading in March can still appear in the September guide. Nothing in the format carries a date of last verification.
Wholesale marketplaces are a different object
A separate family gets confused with the directories because it uses the same words. Faire and Ankorstore are wholesale marketplaces. They are not manufacturer directories, they do not claim to be, and the confusion is expensive because it leads brands to list in the wrong place and buyers to source from the wrong layer.
Faire is a retailer’s buying tool
Faire describes itself as “an online wholesale marketplace with easy-to-use tools designed to help you sell to hundreds of thousands of retailers” 19, and its homepage invites retailers to “shop wholesale online from over 100,000 brands” across categories that include Pets 20. The commercial terms are retailer-facing: low order minimums, 60 days to pay interest free, and free returns on every first order with a brand 20.
Read as a channel, that is a substitute for a trade fair in the independent-retail segment, and it changes the timing of discovery rather than the evidence. Read as a supplier directory, it is a category error. Faire’s unit is a brand with a wholesale catalogue. It does not present itself as identifying factories, and the pages reviewed make no compliance claim about the brands it lists 2019. A buyer who finds a product on Faire has found a seller of goods, not a manufacturer, and the import obligations sit with whoever brings the goods into the market.
The metric trap here is worth stating precisely, because it is the kind that produces bad analysis. Faire’s homepage figure of more than 100,000 brands is a global count with no scope statement on the page 20. Ankorstore’s figure is explicitly European: “over 30,000 brands and 300,000 retailers across 28 countries in Europe” 21. Setting those two numbers beside each other as a comparison of platform scale compares a global brand count with a European one. It is not a like-for-like comparison, and no arithmetic on the pair produces a meaningful share.
Ankorstore is the European mirror, with a European footprint
Ankorstore describes itself as “the largest marketplace for independent retailers and brands in Europe”, a self-description the company attaches to a platform of 30,000 brands and 300,000 retailers in 28 European countries 21. The shape is the same as Faire’s: brands on one side, independent retailers on the other, and a catalogue rather than a factory list in between.
For a supplier, the strategic point is positional. A factory without a brand has nothing to sell on either platform, because both platforms sell catalogues to retailers. A brand owner with a European importer of record can use them to reach independent retail without a booth. Neither platform substitutes for the compliance step, and a listing on either is a commercial arrangement with a retailer-facing promise attached, not a credential.
Databases that record rather than list
The routes that carry real evidentiary weight are the ones where the entry was created by someone other than the seller. There are four worth knowing, and they answer four different questions.
Legal identity: OpenCorporates and VIES
OpenCorporates is an aggregator of official registers, stating coverage of “over 140 government registries”, operation since 2010, 5 million searches monthly, and use by “80% of North America and Europe’s largest banks” 22. Its product is the company record: registration number, jurisdiction, status, address, officers where the register discloses them.
That answers the question that matters at the point of contracting. The name on the purchase order has to be a legal person that exists in a named register on a named date. A storefront name, a trade name and a registered entity name are frequently three different strings, and the mismatch is where advance-payment fraud lives.
VIES does the same job for tax identity inside the EU. The Commission’s service validates “the validity of a VAT number issued by any Member State / Northern Ireland” 23. Two limitations are built into it. It is a validity check on a number, not a check on the business behind it. And the UK left the system: since 1 January 2021 GB VAT numbers can no longer be validated there, and Northern Ireland numbers carry an “XI” prefix 23. A buyer who has not updated a supplier-onboarding checklist since 2020 is still asking for a check that no longer exists.
The marginal value is not that these tools prove a supplier is good. It is that they let a buyer assemble an identity chain: the entity that invoices, the entity holding the export licence, the entity named as shipper, and the entity that owns the brand on the label. When those four coincide, the file is coherent. When they do not, the buyer has found the question to ask.
GS1: the identifier that gets mistaken for an origin claim
GS1 is the standards body behind the barcode, and its record is routinely read as something it is not. Its own pages illustrate why.
The organisation reports local Member Organisations in “over 110 countries” and, on a different page, “over 120 countries”; and “over a million user companies” on one page against “Over two million are part of GS1” on another 2425. Both pairs are first-party statements, published without reconciliation. Neither is a supplier count, and neither measures manufacturing.
The mechanic that matters is allocation. A GTIN is allocated by the brand owner, defined by GS1 as “the organisation that owns the specifications of the trade item, regardless of where and by whom it is manufactured” 25. The prefix therefore identifies the GS1 Member Organisation that issued the number, which reflects where the brand owner registered, not where the goods were made. A pet bed made in Zhejiang and sold under a German brand will carry a German prefix.
Two consequences follow, pointing in opposite directions. A barcode is not an origin claim, and any due diligence that reads a prefix as one is methodologically wrong. But the prefix does identify who owns the specification: a factory quoting a customer’s GTIN is describing itself as a contract manufacturer, and a factory quoting its own prefix is describing itself as a brand owner somewhere. Those are different businesses with different economics.
GS1’s lookup service, Verified by GS1, returns “product, company and location information” for business partners and government entities 26. It is a lookup over data the brand owner supplied; it does not return a factory. Buyers keep reaching for it because a GTIN has become a listing requirement, GS1 being “the only authorised provider of Global Trade Item Numbers (GTINs) for trading on Amazon and leading global retailers” 25. A number a retailer demands becomes, by habit, a number buyers treat as proof. It is proof of a registration.
Customs: the only record of a relationship
Customs data is the highest-resolution commercial evidence available to a buyer who is not yet a customer, because it is the only route in this article that records a transaction between two named parties.
Panjiva, part of S&P Global Market Intelligence, describes “over 2 billion shipment records from 21 countries and 13 million company-to-company relationships”, with history initiated in 2007 and data drawn from “publicly available customs declarations and trade bureau statistics” 27. ImportGenius advertises “16 million+ importers and exporters” and “3.3B+ records from 25+ countries”, with US imports back to 2006 and exports from 2014, priced from USD 199 per month 28. ImportYeti offers a free tier over US customs records, fronted by a live counter that read roughly 878 million records on the access date and updates daily 29.
What the record contains is a shipper, a consignee, an HS code, a weight, a vessel, a port and a date. What it establishes is that a consignment was declared from one named party to another on a date. That is enough to falsify one specific kind of claim, which is the claim of a customer relationship. A factory that says it supplies a named brand either appears as a shipper on that brand’s consignments or it does not.
The limits are as important as the power. The shipper is not necessarily the manufacturer, since trading companies and consolidators appear as shippers routinely. Values are frequently absent or estimated, so unit price cannot be derived. Coverage is not global: 21 countries for one vendor and 25-plus for another is a list, not the world, and both are weighted to the United States. Absence of a record is not absence of a relationship, because goods routed through an intermediary may never show the factory’s name. And a shipment proves a transaction occurred, not that it went well, not that it will recur, and not that the goods met the specification.
The regulator’s list: the one record that can be withdrawn
One primary record in the pet sector can be checked before the first email, is created by an authority with no commercial interest, and can be removed when it stops being true. It exists for pet food and for nothing else in the category.
In the European Union, establishments handling animal by-products are approved or registered by Member State competent authorities, and the lists are public. For non-EU suppliers, the sequence is strict: a country must itself be listed before it may propose establishments, the exporting country’s competent authority enters the establishment data directly into TRACES NT, and the Commission checks the details before publication 30. Pet food plants sit in Section V of the animal by-product lists, described as “Petfood plants (Including plants manufacturing dogchews and flavouring innards)” 30. The legal basis, Regulation (EC) 1069/2009, has applied since 4 March 2011 31.
The removal mechanism is what distinguishes this from every commercial record discussed above. The Commission’s own text states that when a non-EU competent authority identifies that an establishment no longer fulfils the requirements for listing, “it must update its lists and remove the establishment” 30. A listing is therefore a current claim under an obligation, not a historical badge.
The boundary of this record is equally sharp. It covers pet food and chews. It does not cover beds, bowls, carriers, harnesses, toys or grooming tools, because those are consumer products rather than animal by-products, and consumer-product law maintains no approved-supplier register at all. The General Product Safety Regulation, applicable from 13 December 2024, requires an EU-established responsible economic operator, traceability information, a technical file held for ten years and recall capability through Safety Gate, and it creates no certificate and no register of compliant products 32.
The asymmetry is the most consequential fact in this article. The half of a pet brand’s catalogue with the most suppliers has the least public evidence, and the half with a public register carries the heavier compliance load. A buyer sourcing hard goods should stop looking for a register early, because there is not one, and budget for private verification instead.
Three further routes through this material are covered elsewhere in the series: the exhibitor list in The Global Pet Trade Show Calendar: How Buyers Screen a Hall of Manufacturers, the association roster in Trade Associations and Member Directories: The Shortlists Buyers Treat as Vetted, and the customs trail in Reading a Competitor’s Container: How Customs Data Reveals the Factory Behind a Brand.
What triggers the move from a listing to a record
Buyers describe the shift as a change of tool. It is better understood as a change in exposure: the listing is adequate while the buyer is browsing and inadequate the moment being wrong costs something specific.
Five triggers account for most migrations, and each maps to a different record.
The first is the purchase order. A quote is a conversation; a purchase order is a contract with a legal person. At that moment the buyer needs a registration number, a matching entity name, and a bank account in that name. This is where OpenCorporates, a national register or a VIES check enters, and it is the commonest reason a buyer who has been browsing a marketplace suddenly asks for a business licence.
The second is the destination market’s import file. Pet food into the EU requires an establishment on a published list and a health certificate from the exporting country’s competent authority; the list is the gate, checked before a border control post is reached 30. Hard goods require a responsible economic operator and a technical file 32. The buyer’s own compliance obligation, not the supplier’s quality, drives the request, and drives it to a specific document.
The third is a claim that needs falsifying. A supplier says it supplies a named brand. That claim is checkable against customs records in a way that no other claim in the pitch deck is 2729. Buyers who discover this tool tend to overuse it for a quarter and then settle into using it for exactly this purpose.
The fourth is a payment threshold. Below a certain deposit buyers move on a quote. Above it, finance imposes a check, and the check is identity plus accounts, not a directory entry.
The fifth is the category switch. The same supplier may be verifiable in one line and invisible in another, because one line is regulated as food and the other as a consumer product. A factory making both treats will be on an establishment list for the food line and have no public record for the toy line. The buyer’s evidentiary position changes with the SKU, not with the supplier.
The pattern across all five is the same. Nothing about the supplier changed at the moment the buyer moved. What changed was who would bear the consequence of a wrong answer.
What each route proves about a supplier
Against the six questions a sourcing decision turns on, the routes separate cleanly.
| Route | Legal existence | Trading history | Product registration | Capacity | Compliance | Ownership and financials |
|---|---|---|---|---|---|---|
| Marketplace listing | Partly, at paid tiers | Claim only | No | No | No | No |
| Marketplace audit (audited or verified tier) | Yes | Yes, as at audit date | Evidence of certificates seen | Reported, at one visit | Partly, as at audit date | Sometimes, if financial capacity is in scope |
| Vertical directory listing | Claim only | Claim only | No | No | No | No |
| Company registry | Yes, against a named register | Incorporation date only | No | No | No | Entity structure, not solvency |
| Customs records | Shipper and consignee names | Yes, as observed shipments | No | Volume only, and only for covered lanes | No | No |
| GS1 | Brand owner identity | No | No | No | No | No |
| Regulator’s establishment list | Yes, with an authority’s approval | Implied by listing date | Yes, for that activity | No | Yes, for that activity | No |
Two readings are worth keeping. The column a buyer cares most about, capacity, is empty in every open route; it comes only from a site visit, an audit report or a supplier’s own disclosure. And the only route that answers compliance for free is the one restricted to pet food and chews.
The table also explains why buyers layer routes rather than choosing one. A registry check answers who the counterparty is. A customs record answers whether the claimed relationship is real. An establishment list answers whether the plant may ship. An audit answers whether the plant can hold a specification. Four questions, no overlap, and no route that answers more than two.
Where a supplier’s positioning pays back, and where it does not
Suppliers spend on these platforms continuously and rarely measure the return. Spend works where the buyer’s shortlist is formed by search and fails where it is formed by a register.
Positional spend pays back in five situations.
Being findable under the exact term the buyer types, in the buyer’s language, is the cheapest return available, and it is the reason a Europages profile or a Kompass classification entry can be worth more than a banner. A supplier absent from the vocabulary a buyer searches is not evaluated at all.
A registration-based badge pays back, because it converts a self-description into a check. Global Sources’ Verified Manufacturer status, which tests whether the business scope on the licence permits manufacturing, answers the trader-or-factory question that consumes the first two emails of most relationships 5.
An onsite capability report pays back at greater depth, and it is the one paid artefact in the marketplace tier a buyer’s quality team will read 4.
Presence in an upstream directory pays back where the customer is a manufacturer. A co-packer, an ingredient supplier or a packaging converter in WATT’s guide is in front of the population that buys its output, which no pet-specific consumer directory reaches 17.
Owning a GS1 prefix pays back if the strategy is to be a brand. A factory selling under its customers’ GTINs is a contract manufacturer and should price like one; a factory with its own prefix has a product identity it can defend 25.
Positional spend fails in four situations, and they share a structure.
Buying the top of a category fails when the buyer filters before reading. Thomas’s buyer interface lets a buyer filter by certification and business type, and up to 20 suppliers can be contacted in one action 13. A supplier that buys the top of a category it cannot be filtered into has bought placement in front of a buyer who has already excluded it.
Buying a badge fails when the buyer’s compliance file requires a document the badge is not. A Verified subscription does not produce an establishment listing, a health certificate or a technical file 1432. The badge answers a data-quality question; the file asks a regulatory one.
Listing on a wholesale marketplace fails for a factory without a brand, because the platform’s unit is a catalogue sold to retailers, not a production capability sold to brands 2021.
And buying prominence in a directory whose ranking is known to be sold fails hardest of all with the most professional buyers, because they have read the rate card. The suppliers who benefit from paid placement are the ones whose other evidence can survive the disclosure.
The asymmetry that ties this together: marketing spend moves a supplier up a list, and evidence moves a supplier onto a shortlist that is assembled from registers. The first is a purchase. The second is not for sale.
What none of these routes can show
The blind spots are consistent across every platform discussed, and the discipline is to name them rather than let a directory’s formatting imply more than it holds.
No route shows capacity. None of the directories in this article publishes plant capacity, line count or monthly output, and the ones that report capacity report what a supplier told an auditor at a visit. Capacity is confirmed by a site visit, an audit scope that includes it, or an order history that tests it.
No route shows unit cost. Customs values are often absent or estimated, list prices are marketing, and no directory publishes transaction prices 27. Cost is a negotiation, and its inputs are the factory’s own numbers.
No route shows solvency. A registry confirms an entity exists, not that it can pay its suppliers. Financial capacity appears in some audit scopes, as a reported figure, at a date 4. Credit bureaux and trade references are the tools for this, and neither is a directory.
No route closes the identity chain by itself. The listing name, the contracting entity, the licence holder, the shipper and the brand owner of the GTIN are frequently five different strings, and each route resolves one link. The buyer has to assemble the chain, and the assembly is the due diligence.
No route has a reliable expiry. A marketplace badge, a directory entry, a customs record and a GS1 record are all historical statements with no visible decay. Only the regulator’s list carries an obligation to be removed when it stops being true 30, which is why it is worth more than its modest scope suggests.
One problem is structural rather than technical. Where inclusion is free, presence proves only that a form was filled in; where inclusion is paid, absence means only that a supplier declined to spend. Either way the directory cannot be read as a population, and a directory is a map of who markets, not of who makes. Buyers who treat it as the latter are not misreading a page. They are using the wrong instrument, and the error surfaces at the border or at the first failed batch, where it is expensive to correct.
Basis of this analysis
Research was carried out on 15 November 2026. Discovery used DuckDuckGo’s lite endpoint, after which every publisher URL was fetched directly; no figure is taken from a search snippet and no redirect URL is cited.
Every source in the list below is first-party: the platform, standards body, regulator or data vendor describing its own product or rule. That is a deliberate constraint, because the subject is what these platforms say they do, and the alternative (trade press summarising a media kit) adds a layer of error without adding evidence. The cost of the constraint is that this article contains no independent audit of any directory’s supplier population, because none was located in any of the four families.
Pages that could not be retrieved are recorded in the evidence ledger and are not cited for a figure. Global Sources’ homepage returned a bot-protection interstitial, so no Global Sources supplier count is used; its verification page was reachable and is the basis for everything said about its badges. Faire’s June 2026 press release was blocked by the publisher’s protection in an earlier compilation and is not cited; Faire scale figures come from its own live pages.
The WATT rate card is the 2020 edition, the only one located. Prices from it are given with that year. Subscription data products (Panjiva, ImportGenius, ImportYeti’s paid tier) were not purchased; their descriptions are taken from public product pages, and their record counts are treated as vendor claims.
No interview material was used, and no platform was asked to comment. The analysis describes how the routes function and what each one can evidence; it does not recommend a platform.
