The word credentials sounds like one thing. In a window purchase it is at least six, and they are not the same kind of object. One is a government record. One is an audit opinion about a company. One is a laboratory result about a specimen. One is a legal declaration signed by the seller. One is a letter from somebody else's supplier. One is a photograph of a finished building. You cannot verify them with the same move, and the move that works beautifully on the first one does nothing at all on the third.
That is the whole reason this question is hard. Buyers who have been told to "check the credentials" usually perform one check, on the one document that happens to be easy to check, and then treat the other five as done. The rest of this article is about which check belongs to which object, and about the uncomfortable fact that the object you most want to check is the one with no global register behind it.
| Credential | What it actually attests | Who issues it | Can you check it yourself | What it can never tell you |
|---|---|---|---|---|
| Commercial registration | That a legal entity exists, under a name, with a registered scope of business | A government registry | Usually yes, in the registry of that jurisdiction | Whether that entity owns or operates a factory |
| Management system certificate | That an organisation operates a documented system that was audited against a standard | A certification body | Yes, through the issuer, and through a global register if the issuer is accredited | Anything at all about any particular window |
| Product certificate or test report | That one defined specimen, in one defined configuration, met one defined set of requirements | A certification body or a laboratory | Partly, through the issuer only; there is no single global register | Whether your opening, your glass build-up and your sash size are the tested configuration |
| Declaration of performance | That the manufacturer declares values under a named system and takes responsibility for them | The manufacturer itself | No. It is a self-declaration. The reports behind it are checkable, the declaration is not | Who carries the liability in your jurisdiction |
| Supplier authorisation letter | That a named company may buy, machine or assemble a named hardware or system platform | The hardware maker or the system house | Usually yes, by writing to that supplier and asking them to confirm | Whether that platform is rated for the sash you are actually ordering |
| Project references | That something was built, somewhere, by somebody | Nobody. There is no issuer | Partly, if a contactable reference is offered and actually answers | Whether it will be built the same way again, by the same people, out of the same batches |
Read the right-hand column once more. Five of the six rows carry a blind spot that no amount of checking closes. That is not cynicism, it is the shape of the objects. A certificate about a company is not a statement about a window, and no amount of looking at it harder turns it into one.
Start with the sentence that settles more arguments than any other. The international standards organisation states publicly, on its own certification pages, that it does not perform certification or issue certificates, and that it does not permit anyone to use its logo in connection with certification. It goes further: because certification is performed by external bodies, a company or organisation cannot be certified by the standards body at all.
The same source adds that a certified organisation should use the certification body's own logo, not the standards body's logo, and that organisations should not use the standards body's logo to represent their certification status. Which gives you a very cheap first filter: a supplier page that shows the standards body's mark next to the word certified is displaying something that the standards body has asked people not to display. That is not proof of fraud. It is proof that nobody in that marketing department read the page they were borrowing the logo from.
Look at any management system certificate. There is the company that was audited, and there is the company that did the auditing. The second one is the certification body. On most certificates there is also a third party in the form of a small mark: the accreditation body, which is the organisation that assessed the certification body itself.
| Body | What it does | Who watches it | What its absence means |
|---|---|---|---|
| Certification body | Audits the organisation and issues the certificate | The accreditation body | The certificate may be genuine paper with no independent oversight behind it |
| Accreditation body | Formally recognises that the certification body operates according to international standards | Peer evaluation under the international accreditation arrangements | The accreditation may not be recognised outside the economy that granted it |
One honest caveat, and it comes from the standards body itself rather than from us: accreditation is not compulsory, and non-accreditation does not necessarily mean a certification body is not reputable. There are competent unaccredited bodies. There are also accredited ones whose scope does not cover what you are buying. The point of the accreditation mark is not that it proves goodness. It is that it makes the claim checkable, because the accreditation body maintains a register you can look at.
| Link | What you are checking | Where the check happens | What a mismatch actually means |
|---|---|---|---|
| Certified entity | The exact legal name, and every site address named in the scope | The certificate itself, the issuer's own register, and the global database described below | The certificate is about a different company, or about an address that will not make your windows |
| Certification body | Whether it is accredited, and specifically for the standard on your certificate | The accreditation body's register of accredited bodies | Accredited for one standard does not mean accredited for another; scope is granted standard by standard |
| Accreditation body | Whether it is a recognised signatory to the international mutual recognition arrangement | The published membership and signatory lists of the international accreditation forum and its successor arrangement | The recognition stops at the border, which is exactly where an imported window package starts |
The standards body points buyers at a single global database for accredited certification, run in partnership with the International Accreditation Forum and described by that forum as the exclusive global register for accredited certifications. It is the closest thing this industry has to a public truth machine, and it is worth understanding precisely, because its limits are more instructive than its capabilities.
That last bullet is the one to underline. The global register you can actually search is built around management system certification. Your window is not a management system.
Here is the sentence that should change how you file certificates: a certificate is a document, and certification is a status. The document carries dates. The status lives with the issuer, and it can change without the document changing at all.
Management system certification is publicly described as running on a cycle of roughly three years, with surveillance audits at least annually and a full recertification audit before expiry. A certificate can be suspended or withdrawn in the middle of that cycle. When that happens, nothing about the PDF on your desk is altered. It still has its issue date, its expiry date, its number and its logos. The only way to know is to ask the register.
There is a related tell. The forum's digital mark is designed so that if the status of the certificate changes, the mark disappears from the certified entity's website automatically, because the mark is generated rather than pasted. A flat image file of that mark has no such property. A supplier displaying the mark as a static picture is displaying something that the mark was specifically designed not to be: a picture that cannot go stale.
Under the single word certification sit two structurally different activities, governed by two different standards for the bodies performing them. One certifies products. One certifies management systems. They are not interchangeable, regulators treat them as non-interchangeable, and buyers who do not know the difference spend their verification effort on the wrong one.
| Dimension | Product certification | Management system certification |
|---|---|---|
| Standard governing the body | The international standard for bodies certifying products, processes and services | The international standard for bodies providing audit and certification of management systems |
| Subject of the certificate | A named product, model or product family | An organisation, and the scope of its operations |
| Primary evidence | Laboratory test reports on representative samples, plus evaluation of production controls | Audit records and process evidence |
| What invalidates it | A change to the product design or the tested configuration | A change to the organisational scope or to the system itself |
| May the mark go on the product | Yes. That is a defining feature of a product scheme | No. Marks on products are not permitted under a management system scheme |
| In one sentence | The certificate follows the product | The certificate follows the organisation |
Public commentary on this distinction puts it bluntly: a quality management certificate does not certify that any specific product is safe or compliant, it certifies that the processes used to produce products meet the standard's requirements. And a national accreditation body's published guidance adds a second, sharper point that is worth quoting because it hands you a test you can run without any expertise: avoiding the use of the logo on specific products is an indicator that the scheme is not really about product or process certification. If the issuer will not let the mark go on the window, the certificate is not about the window.
Put the last three sections together and the awkward conclusion arrives on its own. The thing a buyer can verify in an afternoon, from another continent, without asking anybody for permission, is the management system certificate. The thing a buyer actually needs is evidence about a window: what configuration was tested, to which method, classified under which scale, by which laboratory, and whether the sash they are ordering is inside that envelope. That second object travels under the product certification rulebook, and there is no equivalent global database you can type a certificate number into.
This is not a reason to stop verifying. It is a reason to change what you ask for. For the product side, the checkable artefact is not a certificate at all. It is a test report with a laboratory name, a standard, a version year, a specimen description and a date, plus a written statement of whether your configuration is the one that was tested.
Even once you have a genuine product certificate, the word covers a wide range. A national accreditation body's published guidance, summarising the international standard on product certification schemes, describes six basic options for how a scheme can be built. At one end sits a scheme where a sample is tested once and nothing happens afterwards. At the other sits a scheme with a validity term tied to ongoing surveillance that includes product testing, process inspection and quality system audits.
| Question to ask | Why it changes what you bought | A weak answer |
|---|---|---|
| Which scheme option applies | It decides whether anything at all happens after the first test | "It is a certified product" |
| What is the surveillance activity | Retesting, factory inspection and system audit are three different things with three different costs | "We are audited regularly" |
| What triggers re-evaluation | If a glass upgrade does not trigger it, the certificate has quietly stopped describing your window | "The series is certified" |
| Who holds the factory production control | Under the system publicly described for ordinary windows, this part is the manufacturer's own | "Everything is tested" |
The same accreditation body guidance draws a line that lands squarely on this industry. It says that a management system certificate leaves the client to evaluate, at the end, whether the producer delivered what was asked for, and that where the client is not capable of defining the product specifications because the product needed is complex and bought once, product certification should be preferred. It gives a one-time purchase of a house or a building as its own example.
A window package for a building is exactly that shape. You will not be able to write the full specification yourself, and you will not be able to test the delivered result yourself. Which means the guidance is telling you to reach for the harder of the two certificates to obtain and, as established above, the harder of the two to verify across a border. That tension is real and it does not have a tidy resolution. It just means the verification work for an imported window package cannot be outsourced to a logo.
If you read one field, read the addresses. Multi-site certification is a well-established arrangement with its own mandatory document, and the rules around it contain one sentence that is worth more to a buyer than any performance number on the page: sampling decides which sites an auditor visits, not which sites the certificate covers. Every site in scope is named. A certificate covering a group carries no conclusion about an address missing from it.
The sampling rule itself is published and simple: the minimum number of sites visited at initial certification is the square root of the number of sites in scope, rounded up, with the central function audited in addition. Across a certification cycle every site must be audited at least once. Applying that published formula ourselves gives the following floors, which are worth seeing as numbers rather than as prose.
| Sites in scope | Minimum sites visited at initial certification | What the buyer should take from it |
|---|---|---|
| 4 | 2 | Half the network is seen in year one. The other half is still certified. |
| 9 | 3 | Two thirds unvisited, fully certified, and no contradiction |
| 16 | 4 | The commonly published worked example, and still only a quarter |
| 25 | 5 | Another published worked example. The sample shrinks as a share of the whole. |
| 100 | 10 | Ninety addresses carry a certificate nobody visited this year |
Two further published cautions belong in this section. First, organisations can have sites that are simply not within the scope of the certification and not listed on it at all; the certificate is not wrong, the site is just outside it. Second, the quieter failure mode described by practitioners is not a bad audit, it is the site that was never on the certificate in the first place, usually because it opened mid-cycle and was never reported. The corrective habit recommended is to read the site list against the addresses on your current contracts, once a year, and raise the difference yourself.
One more structural point, because it decides whether sampling is available at all. Sites are only interchangeable for sampling when they do substantially the same work under the same management system, with a central function that can actually require a change at any site. An office that produces drawings and a plant that extrudes and assembles are not doing substantially the same work. Where the work differs materially, sampling does not reach the different site and it has to be assessed on its own account.
Verification across borders fails most often on names, not on numbers. The name on the certificate, the name on the business registration, the name on the proforma invoice and the name on the bank account should be the same legal entity, spelled the same way. Procurement teams checking group structures routinely ask whether the bidding legal entity is genuinely the certified one, because if one company holds the certificate and a different trading company submits the tender, the buyer has to ask what exactly is covered. The same question belongs on the first page of any enquiry sent to entry doors manufacturers or to window factories: which legal entity will appear on the contract, and is that the entity named on the certificate.
| Document | The name that must appear | What a difference usually means |
|---|---|---|
| Business registration record | The registered legal entity, and whether its permitted business scope includes manufacturing | A scope limited to trading or sales tells you nothing about a press or a paint line |
| The certificate | The certified entity, exactly, plus the site addresses in scope | The certificate may belong to a trading company rather than to the factory |
| The proforma invoice and contract | The contracting entity | Your recourse is against this name, whatever the certificate says |
| The bank account | The same legal entity as the contract | A different account name is the single cheapest signal that the structure is not what it looked like |
| The scope wording | The verbatim scope statement, not a paraphrase | A scope covering trading activities says nothing about fabrication; a scope covering design says something a fabrication-only scope does not |
Sourcing practitioners put this in one line that is worth memorising: a genuine certificate held by the trading company is not the same as a certificate held by the factory, and you should always trace certification to the actual production site. The same author notes that a certificate whose scope only covers trading activities tells you nothing about the shop floor where your parts will be made. That is the entire argument for asking for the site list rather than the certificate summary.
| Signal | Why it is a signal |
|---|---|
| Certification is guaranteed, or offered on a fixed short timetable | No independent third party can guarantee an audit outcome in advance |
| The audit for a complex manufacturing site was done in a day, or entirely by email | Audit duration is one of the things accreditation exists to control |
| A price was quoted without anyone asking headcount, shifts or processes | Audit effort is derived from those inputs; a quote without them is not an audit quote |
| No accreditation mark, or a mark from an accreditor that is not a recognised signatory | The mark is the whole point, because it is what makes the claim checkable |
| No published complaints or appeals process | Both are required of an accredited certification body; their absence is structural, not cosmetic |
| The consultant who built the system and the body that certified it are connected parties | Impartiality is one of the four areas the governing standard exists to enforce |
| The certificate number format is not one the named issuer recognises | A published case involved a competitor's certificate retouched with a new company name, filed for eighteen months and never checked |
Everything above is generic to any manufactured product. Now to the specific artefact. A tilt and turn window is a single sash on a perimeter of hardware that performs two different motions, and the file that describes it has a recognisable shape. The harmonised product standard publicly cited for windows and pedestrian doorsets without fire or smoke characteristics carries a version with an amendment, and that full reference including the amendment year is part of what you should be quoting, because the number alone is the easy half and the year is the half that decides what was tested.
| What you want to see | The document that carries it | The question that makes it useful |
|---|---|---|
| The product standard, with version and amendment | The harmonised standard for windows and doorsets without fire or smoke characteristics | Which edition and which amendment, in writing |
| Air permeability | A test method standard, and a separate classification standard | Was the specimen an opening sash, or a fixed light that never opens |
| Watertightness | A test method standard, and a classification standard running from class one A upward | Which test method, protected or unprotected, and which class number |
| Wind load resistance | A test method standard, and a classification standard | At what pressure, and with which sash dimensions |
| Hardware and safety device strength | A dedicated test standard for safety devices, and a soft body impact standard | Whether the tested device is the one on your bill of materials |
| Thermal transmittance | A calculation standard series | Calculated or measured, and for which reference size |
| Acoustic performance | A laboratory measurement standard series | Measured with which glass build-up |
| Burglar resistance, if claimed | A resistance classification standard | Class, and whether the locking hardware is the tested set |
| Fire and smoke, if required | A separate product standard, outside the window standard above | Whether the project needs it at all, which is a question for your authority, not for us |
For orientation, one classification ladder is widely reproduced in public consultancy material: the watertightness classes run in steps from a class with no numerical requirement, then upward through a sequence of pressure differences, with the upper classes described as suited to commercial buildings and exposed locations, and the highest numbered class to high-rise buildings and coastal areas. The same public material notes that the test is performed either with the specimen protected or unprotected, and that the two are not the same result. Treat that ladder as a description of one publicly published table and verify it against the current edition before you rely on a class number.
Then the part that surprises people. Public summaries of the conformity system for ordinary windows and doors without fire or smoke characteristics describe initial type testing being carried out by a notified body or a laboratory accepted by it, with factory production control managed by the manufacturer itself. Read that twice. The part that decides whether the hundredth window matches the first one is not the part that anyone outside the factory can read.
A tilt and turn sash does two motions on one frame, and the mechanism that makes that possible is made by somebody who is not the window factory. That produces a credential with an unusual property: it belongs to a third party, and it is usually easier to verify than the window factory's own paperwork, because you can write to the hardware manufacturer and ask whether a named company is authorised and whether your sash dimensions fall inside the published application range.
It is also, in practice, the ceiling. The largest sash a given platform will carry is published by the platform owner as a function of sash dimensions and glass weight, not by the factory assembling it. We have written about that ceiling elsewhere, so the point here is narrow: when you are verifying credentials, ask for the hardware supplier's letter and for the application range table, and check your own numbers against it yourself. It is the one credential in this article that you can verify without anybody's permission and without any expertise.
For a sense of what this looks like in practice on our own side, see the Projects pages and the tilt and turn category, and then ask us for the documents behind them rather than the photographs.
We are the manufacturer writing this, so the honest section is the one about us. The pattern to notice is that several of our published numbers are perfectly real and perfectly useless as credentials, because they measure a different property from the one a buyer assumes.
| Item | What we can say in writing | Why the gap is there |
|---|---|---|
| Production sites | Two centres, in Foshan and Zhaoqing, Guangdong | Two addresses is precisely the case where the site list, not the certificate summary, is the operative document. Ask for it. |
| Thermal values we publish | Values as low as 1.3, 1.6 and 1.7 depending on series | They are system values, not whole-window values, and not values produced under the North American rating programme. Several series publish none at all. |
| Salt spray hours and shear strength | 240 hours and 5,000 newtons, on one series | They describe a coating and its substrate, and the composite bond between strip and aluminium. Neither is an air, water or wind class, and neither transfers to another series. |
| Load figures on sliding products | 200, 220 and 400 kilograms, on three different products | They are three different measurements. Sorting them into one descending column and reading it as a ranking is the most common way this catalogue is misread. |
| One series with an incomplete published specification | That the published detail is incomplete | We would rather say that than let you build a schedule on a placeholder |
| Air, water and wind classification | The report for the configuration you name, where we hold one | Classification attaches to a specimen, not to a series name. We do not publish it across the catalogue. |
| Hardware | Which platforms we assemble with | The application range that limits your sash belongs to the platform owner, and is their document to publish |
| Installation | Drawings, profiles, fixings guidance and sample pieces | We do not install. The opening and the perimeter seal are yours and your contractor's, and no factory certificate covers them. |
| Project and jurisdictional approval | Evidence, drawings, declarations and test pieces | The applicant is you, or your local representative. We cannot hold that approval and we do not pretend to. |
| Categories with no products in them | That they are empty | Our folding door category page exists with no products on it, and a china swing door search on our site reaches the same kind of page. We would rather you find that out from us than after the deposit. |
| Origin | Where the goods are made and shipped from | Country of origin is a shipping and customs fact. It is not a performance statement and it is not a credential. |
| What we will not do | Publish a score, a rating, or anybody else's certificate number | Reproducing other people's identifiers is how stale and misattributed records spread. If you want to verify, verify against the issuer. |
The practical output of all this is a short request. Send it once, before the deposit, and read the answers rather than the attachments. Note the middle column: for a buyer comparing several suppliers, the field that is most often left blank is also the field that decides the whole purchase.
| Field | A good answer looks like | If it comes back blank |
|---|---|---|
| Exact legal name of the certified entity | One name, matching the contract and the bank account | You do not know who you are paying |
| Every site address in scope | A list, naming the address that will make your order | Your windows may come from an address the certificate says nothing about |
| Certification body and accreditation body | Both named, both searchable | You cannot run the second and third links of the chain |
| Certificate identifier | A number you can type into the issuer's own register | You are looking at a picture, not at a status |
| Standard, including year and version | A full reference, not a bare number | You do not know which edition was tested against |
| Scope statement, verbatim | The exact wording, including whether fabrication is in it | A scope about trading tells you nothing about the plant |
| Cycle dates and most recent surveillance | Issue date, expiry, and the date of the last surveillance visit | You have a document with no way to tell whether the status still holds |
| Product standard and test methods for your configuration | Named methods and named classification standards | You are comparing a company certificate as if it were a window result |
| Specimen description | Sash dimensions, glass build-up, opening configuration | You cannot tell whether your window is the tested one, which is the entire question |
| Laboratory name | A named laboratory, with its own accreditation status | There is no way to check the measurement, only the claim |
| Hardware platform and application range | The platform owner's letter, and the range table | Nobody has checked that your sash is inside the mechanism's ceiling |
| The entity that signs and is paid | One name, repeated on contract, invoice and account | Your recourse is against an entity you have not identified |
| Ask | Sounds reasonable, means nothing |
|---|---|
| Which address will make my order | "All our production is certified" |
| Which accreditation body, and is it a recognised signatory | "We are fully certified to international standards" |
| What is the certificate number | "We can send the certificate" |
| What is the scope, word for word | "Manufacture of windows and doors" |
| When was the last surveillance visit | "The certificate is valid" |
| Which edition and amendment of the product standard | "It is tested to the European standard" |
| Was the specimen an opening sash or a fixed light | "The series is tested" |
| Which laboratory, and what is its status | "A third-party laboratory" |
| Which hardware platform, and what is its application range | "German hardware" |
| Who signs the perimeter seal | "Installation is straightforward" |
| Situation | Why the certificate cannot help |
|---|---|
| Your opening dimensions are not fixed yet | Classification belongs to a specimen. Nothing can be matched to a specimen that does not exist. |
| You need a local approval with a named applicant | That approval attaches to a project and an applicant. A factory cannot hold it for you. |
| You want one party to measure, install and guarantee | Factory-direct supply separates those responsibilities. If you want them fused, buy locally. |
| You need a spare part this week | The certificate that helps you in year eight is a parts list and a batch record, not a management system certificate. |
| The product you want is not in the supplier's catalogue | Certification follows what is actually made. A category page with nothing on it certifies nothing. |
| You want a number you can paste into a drawing | Thresholds belong to your authority having jurisdiction. Ask them in writing, and do not take them from any article, including this one. |
No. The standards body states publicly that it does not perform certification and does not issue certificates, and that a company cannot be certified by it. It also asks organisations not to use its logo to represent certification status. If a supplier page shows that logo next to a certified claim, the useful move is not to accuse anyone of fraud but to ask which certification body actually issued the certificate, and then to check that body.
Not for everything. There is a global register for accredited certification, described by the international accreditation forum as the exclusive global register of its kind, and it is searched by company name or certificate identifier. Its own guidance says it shows standard, scope, status, certification body and accreditation body, and no dates, and that it is built around certificates issued under the management system conformity standard. Product certificates travel under a different rulebook and do not have an equivalent single global register, which is why the product side has to be checked against the issuer directly.
No, and this is the single most common mistake. A certificate is a document; certification is a status held by the issuer. The document carries dates and the status can change without the document changing. Management system certification runs on a cycle of roughly three years with surveillance at least annually, and it can be suspended or withdrawn mid-cycle while the PDF keeps circulating. Check the register, not the file.
Only if that address is named. Published guidance on multi-site certification says plainly that sampling decides which sites an auditor visits, not which sites the certificate covers, and that a certificate covering a group carries no conclusion about an address missing from it. Sites can also sit outside the scope entirely. Ask for the site list and read your own factory address against it.
Because a copied identifier goes stale immediately and cannot be traced, and because reproducing somebody else's record is exactly how misattributed numbers spread through an industry. We are also a manufacturer writing about our own category, and publishing or repeating third-party scores and identifiers would make this page worse, not better. Verify against the issuer.
The test report, and specifically the specimen description inside it. Everything else in this article is about a company. The specimen description is about a window, and it is the only place where the words tilt and turn, a sash size and a glass build-up appear together on a page that somebody outside the factory signed. If the specimen does not match what you are ordering, no certificate above it matters.