A plush keychain looks like a small plush toy with a ring on top. In manufacturing terms it is two products from two supply chains, joined by a piece of webbing.
The plush body is the part a plush factory controls completely: pattern, fabric, filling, embroidery, stitching. The hardware — split ring, lobster clasp, swivel, chain — is bought in from a metal component supplier. That half is where keychains actually fail, and it is the half that most buyers never specify beyond “a keyring”.
Two supply chains in one product
Ask a plush factory for the metal spec and you often get a photograph of a sample. Ask a hardware supplier for the plush spec and you get a blank look. The finished keychain is a systems integration problem wearing a cute face.
The practical consequence: your compliance exposure on a keychain is mostly not in the building that makes it. The factory can control the seam strength and the fill weight. It cannot control what alloy the plating bath contained unless somebody asked, in writing, before the order.
So the specification has to cover both halves, and most of the attention belongs on the half that is not plush.
Nickel release: the test that catches people
The governing rule in the EU is REACH Annex XVII, entry 27. For articles intended to come into direct and prolonged skin contact, nickel release must not exceed 0.5 µg/cm²/week.
Three things about that sentence matter more than the number.
It is release, not content. There is no pass or fail for “does this contain nickel”. The question is how much migrates out of the surface under a defined test. A part can contain nickel and pass, because it is sealed under chromium; another can look clean and fail because the plating is thin and porous at an edge.
It is measured after wear. The reference method is EN 1811 for release, preceded by EN 12472, which simulates wear and corrosion. Fresh, perfect samples are not what gets tested, and a hardware sample that arrives polished from a supplier is not evidence of anything.
Skin contact is the trigger, not the word “toy”. A plush keychain hangs from a bag, a belt loop, or a pocket. That is prolonged contact, which is exactly the scenario the rule describes. The nickel requirement follows the contact, regardless of how the other standards are applied.
Why electroplating is the usual culprit
Electroplated zinc alloy is the default for bag hardware: cheap, easy to cast into a lobster clasp, and it takes a bright finish. The standard plating stack is copper, then nickel, then chromium. The nickel layer is what gives the chromium its brightness and adhesion, and it is also what the release test finds when the stack is thin or worn through.
The alternatives, in order of what they cost:
| Hardware approach | Nickel release risk | Trade-off |
|---|---|---|
| Plated zinc alloy, thin stack | High, especially at edges and wear points | Cheapest, brightest, most common |
| Plated zinc alloy, heavy sealed stack | Low if the seal is intact and the sample is wear-tested | Cost, and process control you cannot see |
| Stainless steel 304 or 316 | Usually well below the limit, but it does contain nickel and is still a release-tested material | Higher unit cost, harder to cast fine shapes |
| PVD-coated or painted hardware | Low, but a coating that chips becomes a different problem | Colour range is narrower |
A supplier claiming “nickel free” without a report is describing an intention. The document that matters names EN 1811 and EN 12472 and gives a measured value.
The other metals
Nickel gets the attention because of contact dermatitis, but it is not the only line on a hardware BOM.
- Lead. Under CPSIA, accessible substrate is limited to 100 ppm, and surface coatings to 90 ppm. Cheap zinc castings are a classic place to find it, because lead improves machinability.
- Cadmium. Restricted under REACH and covered by migration testing on the toy side; it turns up in low-grade plating and in solders.
- Migration generally. If the keychain is a toy, the hardware counts as toy material and falls under the migration test for elements alongside the fabric and the eyes.
This is why a hardware specification is not a detail. It is a purchasing decision with a test attached.
The attachment: pull-out and small parts
The second failure mode is mechanical, and it is the one a customer notices in a week.
A keychain hangs from a ring, and the ring takes the entire weight of the toy plus whatever it snags on. The points that fail are predictable:
- The webbing or loop where the hardware passes through the plush. This is a seam and a fabric strength question, and it belongs to the factory.
- The split ring gap. A ring that opens under load releases the whole toy, and it is also a small part in its own right.
- The clasp spring. A lobster clasp whose gate does not close fully is a snap hook with a decorative purpose.
Treat the assembled keychain as one component and subject it to the same tension test you would apply to a safety eye. If the ring pulls free below the threshold for the age band, the product fails regardless of how well the plush is made.
And this is where age grading bites. The standard small parts cylinder is 31.7 mm in diameter and 57.1 mm deep. A split ring, a clasp, or a swivel passes through it easily. Unless the hardware has been assessed for the under-3 band, a plush keychain ships as 3 and up, and that has to be on the label, the packaging and the listing — not just in the factory’s head.
Is a plush keychain legally a toy?
This question decides your entire test plan, and buyers get it wrong in both directions.
A keychain marketed to children under 14, sold in a toy channel, or shaped as a plaything is a toy, and the toy safety standards apply in full — mechanical, flammability, migration.
A keychain sold as a fashion accessory to adults may fall outside that scope. That is not a loophole, it is a classification, and it has to be defensible from the packaging and the listing rather than from an email after a customs query.
Two practical notes:
- The nickel rule survives either classification. It attaches to skin contact, not to toy status.
- Classification is the buyer’s risk, not the factory’s. The importer puts the product on the market; the factory builds to the specification it was given. Specify which regime you are selling under, in writing.
This is the same principle that runs through what each certificate actually proves, applied to a product where the answer is unusually easy to get wrong.
Weight and fill, on a bag
A keychain is carried, not hugged, and that changes the fill decision.
A standing character is stuffed to hold a silhouette. A keychain that weighs 90 grams drags a tote bag sideways and swings into things. Under-filling by comparison is normal, and the goal is a body that keeps its shape without committing the hardware to load it was not specified for.
The practical spec is a gram target for the finished body, agreed at sampling, plus the rule that the hardware carries only the toy’s own weight during normal use.
Size works the other way. Below about 10 cm, embroidery detail becomes the limiting factor before the shape does: eyes and features that are legible on a 20 cm character turn into a smudge on a 9 cm one, and the fix is a simpler design rather than a finer stitch.
What to put on the purchase order
| Line | What to write | Why |
|---|---|---|
| Hardware material | Alloy and plating stack, not just “metal keyring” | Decides the nickel route |
| Nickel release | EN 1811 after EN 12472, with the measured value in the report | Release, not content |
| Lead and cadmium | Limits per destination market | Found in cheap castings |
| Attachment strength | Tension test value for the assembled keychain | The ring pulls out before the seam does |
| Age grade | 3+ unless the hardware is assessed for under-3 | The label has to match the assessment |
| Fill weight | Gram target for the finished body | Weight on a bag is a design input |
| Toy or accessory | State the classification you are selling under | It changes the test scope |
Where this stops
None of this certifies a shipment. The tests have to be run on the actual hardware lot you are buying, and a report from a previous order is evidence about a previous order. Plating baths change, alloys change, and a hardware vendor that passed last year is not a guarantee about this year.
What the specification buys you is a named, testable claim instead of a photograph of a sample — and that is the difference between a keychain you can sell in the EU and one that gets held at the border.
What we do on a keychain order
| Item | Detail |
|---|---|
| Plush body | 10 to 15 cm typical, with the fill weight specified in grams at sampling |
| Hardware | Specified by alloy and plating on the BOM, not by sample photograph; sourcing named in the quotation |
| Metal testing | Carried by third-party accredited laboratories where the destination market requires it |
| Attachment | Tension tested as part of the pull and torque regime on the assembled keychain |
| Age grading | 3+ by default, and the label reflects the assessment rather than the intent |
| Inspection | 100% needle detection and AQL 2.5 on the finished keychain |
Related reading
- Age grading a plush toy: the small parts rules behind a 0+ or 3+ label
- Lead and phthalates: which components actually need testing
- What each plush toy certificate proves, and what it does not
- Custom plush toy manufacturing: standards, specifications, MOQ and lead times
- Panda plush keychain at 15 cm and chestnut bear plush at 10 cm: two ends of the size band where hardware mass matters
- The full keychain range: every style with its size, fabric and MOQ
Flechazo manufactures OEM/ODM plush keychains and bag charms from 10 cm, with hardware specified by alloy and plating on the BOM and metal testing carried by third-party accredited laboratories. MOQ from 100 pieces per design, prototyping in 5 to 7 working days. Send the character and the destination market, and the quotation will name the hardware and the test it carries.