The artwork looks like the least technical document in the project, which is exactly why it is left until the last minute. Here is what the reviewer is actually comparing, and what the printed label has to match.
You have sent samples, a manual and a BOM. The laboratory replies with one more request: the label artwork, or the nameplate drawing. It looks like the least technical document in the whole project, which is exactly why people leave it until the cartons are already booked.
There is a reason the reviewer wants it. A test report describes one specific product configuration, and the label is the only part of that configuration that ships on every single unit and makes claims to whoever opens the box. If the report and the printed label say different things, they no longer describe the same product.
When the artwork arrives it gets put next to two other documents: the model list the report will cover, and the ratings and configuration that were actually tested. The marking, the warnings and the responsible-party line are then read against the manual and the declaration.
Anything that disagrees is either corrected on the artwork, or - if it cannot be corrected - becomes a scope question for the test programme. Catching it at artwork stage costs a file edit. Catching it after the labels are printed costs a reprint, and if the mismatch is in the model or the ratings, it can cost a retest.
| Element on the label | What the reviewer checks |
|---|---|
| Type or model designation | Is the printed model code inside the model list the report covers? A marketing code that never appeared on the report is not covered by it. |
| Electrical ratings | Input voltage and frequency, and the current or power figure, must match the ratings the product was tested at. On adapters and power supplies the output rating matters too, because it sets part of the test load. |
| Manufacturer or responsible party | A name and a postal address. Where the manufacturer sits outside the market, that market's importer or responsible person usually has to be identified on the product or its packaging as well. |
| Market marking | The mark itself, at the required size and proportions, plus the notified body number where a notified body was involved. |
| Equipment identifier and compliance statement | For the US, the equipment identifier and the Part 15 statement, on the product or, where the labelling rules allow it, in the manual. |
| Safety symbols | Class II double square, IP rating, indoor-use and caution symbols. Every one of them is a claim the file has to support. |
| Warnings and language | The warnings the standard or the manual require, in the language of the market where the product is sold. |
| Batch or serial area | Whether traceability is on the product or the packaging, and whether the print method can actually hold it. |
It is a short list, and every line on it is a place where a design file and a test file can drift apart. Three of them cause most of the problems we see: the model code, the ratings, and the address block.
Two mechanical questions arrive with the artwork. Both are easy to settle before printing and expensive after.
Position. Marking normally has to sit on the product itself, visible without tools, or on a nameplate fixed to it. Moving it to the packaging or the manual is only allowed in the cases the rules name - for example where the product is too small to carry it. For the EU, the general principles sit in Regulation (EC) No 765/2008 Article 30: the mark goes on the product or its data plate, and only where that is not possible on the packaging and accompanying documents, but in every case it has to be visible, legible, indelible, at least 5 mm high and fixed in proportion to the design. Electronic labelling is a separate route that exists for devices with their own screen and comes with conditions on how the information is reached; a device without a display normally still carries a physical label.
Durability. Most safety standards that require marking also test that the marking survives handling. In the household appliance standard IEC 60335-1, for example, the marking is rubbed with a cloth soaked in water and then with petroleum spirit, and it has to stay legible afterwards. That is why the reviewer asks what the label is printed on and how - a paper sticker that peels, smudges or fades will not pass, however good the design file looks. If the label sits somewhere a user touches daily, settle this at artwork stage.
The artwork carries a model code the report does not name. The report lists a base model with a few variants; the retail artwork carries the marketing name printed for each market. Neither is wrong on its own, but a report that covers model A does not cover a product labelled model B. If the same electronics are sold under several codes, those codes belong in the report's model list from the start, before the test is scheduled.
The printed rating is not the tested rating. An adapter tested at one output rating with artwork printed at a different one is a mismatch, even when the hardware is identical. The same thing happens with the input range: a product tested across a wide input range but labelled at a single voltage undersells itself, and a label claiming a range the sample was never tested across is a claim with nothing behind it.
The artwork makes a claim nobody tested. "Waterproof", an IP rating, or a Class II double square on a product that was tested as Class I. Marking is not decoration - each symbol points back to a requirement in the file. Either the test is added or the claim comes off the artwork, and that is a far cheaper decision before the printing plate is made.
The responsible party changed after the artwork was approved. A distributor switches importer, or a brand owner decides the address on the label should be theirs rather than the factory's. The test result is untouched, so this is a label change rather than a retest - but every printed label and every carton in stock now names the wrong entity. Agreeing who signs the declaration and appears on the label before print is one of the cheapest risk controls in the whole project.
| Change | Usually needs |
|---|---|
| Fixing a typo, spacing, or the address block | New artwork and a note to the file. No testing. |
| Adding a model code that is already inside the tested configuration | A model list amendment, sometimes with a short engineering review. |
| Adding a model with a different radio module, hardware revision or rating | Evaluation of the difference, and often partial or full retesting. |
| Changing the declared rating on the label | A check against the test conditions. Ratings decide part of what was measured, so this is not always free. |
| Removing a required warning from the artwork | A label correction - and a reminder that the warning is required, not optional. |
| Adding a claim the file does not support, such as an IP rating | Testing that claim, or removing it from the artwork. |
The pattern is straightforward: anything that changes what the product is touches the test file. Anything that only changes how the same product is described usually does not. The awkward cases sit in the middle, which is why the artwork is worth reviewing while there is still time to choose.
Artwork usually exists in two versions, and both are useful.
Before you send either version, confirm four things internally: the model list, the ratings, who is the responsible party in each market, and which warnings the standard or the manual require. Those four answers are what the artwork is measured against, and they are much easier to agree as a team than to reconstruct afterwards.
Label review sits inside the document work on every project: matching the artwork to the model list and the tested ratings, checking that the marking, symbols and address block are complete for the markets you ship to, and flagging anything that would need a test rather than an edit. If the label raises a scope question, it gets answered before the report is written, not after. You can see how that fits into a normal project in our testing and certification scope and how we run a project. The rest of the document set is covered in what a laboratory needs from you, the manual specifically in why the lab asks for the manual, and the usual test scope by product type on our product category pages.
Send us the label artwork you plan to print, together with the model list and the ratings you have tested. We will mark up what is missing, what does not match the file, and what has to be corrected before the artwork is released.
Because the report covers one specific configuration, and the printed label is what ships on every unit. The model code, ratings, mark and address block on the artwork have to match what was tested. Checking it before printing is the cheapest place to catch a mismatch.
Yes. A report that lists model A does not cover a product labelled model B, even when the hardware is identical. If you sell the same product under several codes or marketing names, all of them should appear in the report's model list from the start.
Often yes, if the product itself is not changing. Correcting an address, a typo or a spacing issue, or adding a model code already inside the tested configuration, is a document amendment. Changing a declared rating, adding a radio variant or adding a new claim is a scope question and can require testing.
Yes. The general principles in Regulation (EC) No 765/2008 Article 30 require the mark to be visible, legible and indelible, at least 5 mm high, and affixed in proportion to the design. It goes on the product or its data plate, and only where that is not possible on the packaging and accompanying documents.
FCC rules do allow electronic labelling for devices with their own screen, under the conditions set out in the labelling rules. A device without a display normally still needs the physical label, and the compliance statement can move to the user manual only where the rules permit it.
The report stops being evidence for the product as shipped. Goods can be held while a reprint or a retest is arranged, and a listing can be pulled while the documents are corrected. Both are avoidable by checking the artwork against the tested configuration before print.