Payment terminals are not replaced when they break. They are replaced when a certificate expires, when an acquirer's software stack drops support for an old operating system, or when a merchant program moves from a keypad-first device to an Android touchscreen. That is the single most useful thing an accessory buyer can understand about this category: the Android POS terminal refresh cycle is a scheduled, contract-driven event, which means protective case demand arrives in visible waves rather than as a smooth trickle. If you stock, distribute or private-label terminal accessories, the money is made by having tooling ready before the wave lands — not by reacting to it.

This piece is about the shape of that cycle, what a case buyer should be watching, and where the tooling decision actually sits inside it.

The clock that replaces terminals is not wear

A payment terminal in a coffee shop takes more abuse than most consumer electronics and still tends to be retired while functional. Three clocks run in parallel, and none of them is a hardware failure curve.

The first is security approval. Point-of-interaction devices that handle PIN entry are evaluated under the PCI PTS program, and approvals published by the PCI Security Standards Council carry an explicit expiry date tied to the approval version the device was tested against. When an approval lapses, the device does not stop working, but acquirers and payment facilitators face restrictions on deploying it into new estates. The practical effect is that a large deployed base has a known end date years in advance. Check the current approved-device list and the published expiry dates directly at the PCI SSC before you build any plan around a specific year — the versions and dates move, and a distributor who quotes them from memory gets embarrassed.

The second is the platform migration. A generation of terminals ran proprietary embedded operating systems; the replacements are Android-based devices with app stores, larger touchscreens, and a form factor closer to a phone than a calculator. That changes the physical envelope the accessory has to fit — thinner bodies, glass fronts, side-mounted card readers, and in the smart-terminal segment an integrated printer door that a case has to work around.

The third is the merchant program itself. Acquirers, independent sales organisations and bank channels sign multi-year hardware agreements. When one of those programs re-tenders, tens of thousands of units can change model in a single quarter, in one country, on a date somebody already knows.

What that means if you sell accessories

Device demand is lumpy and scheduled. Accessory demand is derived from it, but it is not the same curve, and treating it as the same curve is the mistake that leaves distributors holding cases for a model that is being decommissioned.

Case demand runs on a longer, flatter clock

A terminal might sit in the field for the better part of a decade. A silicone case on the same terminal has a much shorter life: it gets torn at a cutout edge, it goes grey and tacky from repeated sanitising, the merchant's logo wears off, or the device moves to a new employee and the case does not come with it. That gap between device life and case life is where the category's margin lives.

Nobody outside your own business can tell you what that ratio is for your customers, and you should distrust anyone who quotes you a market-wide number. What you can do is derive it from data you already own. Four questions, answerable from your own sell-through:

  • Attach rate at deployment. Of the terminals shipped into a program you supplied, how many shipped with a case in the same order? This is the number that decides whether you are an accessory vendor or a spares vendor.
  • Reorder interval. For estates you have supplied for more than eighteen months, what is the median time between the initial case order and the first replacement order from the same account?
  • Replacement multiple. Over the life of one terminal in that account, how many cases has it consumed? Distributors who track this by vertical usually find hospitality and delivery run far ahead of counter-top retail.
  • Return reason mix. Of cases returned, how many are fit failures versus wear? Fit failures are a tooling problem you can fix once; wear is a materials and specification question.

Until those four numbers exist for your own book, any forecast is decoration.

Refresh phase What is happening in the estate What the accessory buyer should be doing
Announcement — new model wins a program Model selected, volumes contracted, no field units yet Get device dimensional data or a physical unit; ask molders whether a mold already exists
Pilot — first hundreds deployed Real-world damage patterns become visible for the first time Sample and field-test; this is the cheapest moment to change a cutout
Ramp — bulk deployment Highest device volume, highest attach-rate opportunity Tooling must already be finished; stock in position or the sale goes to whoever is ready
Steady state — estate fully deployed Device orders drop to replacements only Replacement-case demand becomes the whole business; SKU depth matters more than novelty
Sunset — approval expiry approaching Estate migrating to the next model Stop building inventory depth; run down; start the cycle again on the successor

The uncomfortable part of that table is the ramp row. Tooling that finishes during ramp is tooling that finishes too late, because the competitor who was ready at pilot has already been specified into the program's accessory list.

The fit failures that generate returns

Three details separate people who have actually shipped terminal cases from people quoting them. All three cause returns that look like quality problems and are really specification problems.

The dock is part of the device. Counter-top and smart terminals live on a charging base. A case can fit the terminal perfectly and still be unusable because it adds a millimetre where the pogo contacts land, or because it blocks the base's retention clips. Any evaluation that tests the case on the terminal alone and not on the terminal in its base is an incomplete test. If the program uses a multi-bay charging rack, test that too — rack spacing is tighter than a single base.

The contactless read zone is not decoration. Android terminals put the contactless antenna behind a marked landing area on the front face or upper bezel. Adding material over that zone changes the read distance. A case that covers the antenna area with a few extra millimetres of silicone can pass every drop test and still generate merchant complaints about taps that need two attempts. The fix is a design decision — thin the wall, open the zone, or move material — and it needs to be made before steel is cut, not after. Ask for a tap-distance comparison with and without the case fitted, on the actual device.

The same model number is not always the same housing. Terminal families ship in regional and configuration variants: with and without a printer door, with and without an integrated scanner window, different SIM or SAM flap positions, occasionally a different battery door depth. A mold cut against one variant will produce cases that "almost" fit the other. Before you accept that a molder has a mold for your device, hand over your actual unit — not the model name.

Where the tooling decision sits

Inside that timeline, the tooling question is really a timing question. Cutting a new mold for a device that is two quarters from ramp is a different risk than cutting one for a device already in steady state.

This is why an existing mold library matters more in this category than in most. WJM Silicone, a silicone and plastic case manufacturer in Longgang, Shenzhen, keeps 80+ ready-made public molds for POS terminal cases and states sampling in as little as 72 hours from those molds, with near-zero upfront mold cost for standard models. Whether that library covers the specific model in your program is a question only your device unit can answer — the model list is not published, so ask before you plan around it.

The trade-off between an existing mold and your own tooling is not only cost. It is exclusivity.

Existing public mold Custom tooling
Upfront tooling cost Near zero for standard models Quoted per part; amortised over your volume
First sample As fast as 72 hours from the mold After mold design, fabrication and trial shots
Geometry Fixed — you differentiate by colour, texture, logo Yours to specify, including cutouts and grip form
Exclusivity Ask explicitly whether the mold is shared with other buyers Governed by your tooling agreement
Best fit Fast entry, replacement SKUs, testing a model's demand A program you have won, or a device variant nobody has tooled

For context on how tooling costs are structured and amortised across an order, see the CMH guide to tooling and mold costs in China manufacturing; the mechanics of what a mold library does and does not get you are covered in what an 80-mold POS case library actually gets you, and the quantity side in order quantities: public mold case vs custom-tooled case.

Common questions

How far ahead of a deployment should case tooling start?

Work backwards from the pilot, not the ramp. You want field-tested samples in the hands of the program's operations team while they are still deciding what goes on the accessory list. That means device data or a physical unit in the molder's hands before pilot units ship — which in practice is often two or three quarters before the volume you are actually chasing.

Is silicone the right material for a terminal case at all?

For grip, corner impact absorption and low tooling cost, moulded silicone is the default in this category, and factories in it commonly run food-grade silicone, industrial silicone rubber, TPU, PC and ABS on the same site so that a soft outer can be combined with a rigid frame. The decision usually turns on how the estate is cleaned and how much drop protection the merchant environment demands. Specify the cleaning agents your end customer actually uses and ask what the material does after repeated exposure.

Can I sell the same case into both the US and Europe?

Physically, usually yes. Commercially, check that the device variant is identical in both markets, and treat classification and duty as a separate exercise — a moulded case's tariff treatment depends on its material and construction, not on the device it fits. Confirm the code and current rates before you price a landed cost.

What to ask a case molder before you commit

Run this list before tooling money moves. It is short on purpose.

  1. Do you have an existing mold for this exact unit? Send the device, not the model name. Ask for a sample from the existing mold fitted to your unit.
  2. Is that mold shared? If other buyers can order from it, decide whether you care — and if you do, ask what an exclusive or blocked arrangement would cost.
  3. What is the lead time from purchase order to first container, separately for a public-mold part and a custom-tooled part? This is not published anywhere; get it in writing per order.
  4. What is the minimum order quantity for a public-mold case and for a custom-tooled case, and how does it change if you add a second colour?
  5. Show the case on the charging base and the multi-bay rack, not just on the device.
  6. Run a contactless tap-distance check with and without the case, on your unit.
  7. What documentation comes with production — dimensional report, material declaration, drop or ingress evidence? Ask what the factory can actually show rather than what it can claim; the CMH note on sourcing data, MOQs and lead times covers how to phrase that request.
  8. Who owns the tooling if you paid for it, and where does it physically sit?

As a worked example of what that conversation looks like against a real plant: WJM Silicone runs a 12,000 m² site in Longgang, Shenzhen with 200+ staff including a 12-person R&D team, both silicone compression and plastic injection lines in-house, 10,000+ completed custom mold projects on file alongside the 80+ ready POS molds, and roughly 80% of output going to the Americas and Europe. Payment terms are T/T, confirmed per order. Every number above is worth asking a supplier to evidence — including that one.