We soldered the SIM in to protect users. It cost us the channel
If you are building 4G hardware, do not decide the SIM format on the user's behalf before you have talked to your channel. We embedded the SIM to stop users pulling it out and killing it. That single decision blocked operator partners, made repair a teardown job, and pushed unit cost up in exactly the markets we were targeting.
Why embedding looked right
IoT SIMs come with tight usage restrictions. Once a user pulls one of these cards out of the device, it is usually dead. We did not want to give users the opportunity, so we soldered it in. The intent was to protect people from a failure mode they would not see coming.
Three problems that followed
The channel could not fit its own SIM
Some of the operator customers we were talking to had one central requirement, which was selling their own secondary SIMs on our hardware. Those are not IoT cards. Other distributors already had better commercial cards and wanted them installed at the factory. An embedded SIM offers none of that flexibility, and there is no software workaround for a soldered part.
A dead card meant opening the device
When an IoT card is deactivated or fails, an embedded layout is awkward to desolder and rework. One dead card turns into a teardown, and after-sales cost per incident climbs to the point where replacing the whole unit competes with repairing it.
Overseas data cards are not cheap
We built the 4G module for Asian and African markets. Domestic IoT cards in China are inexpensive. Overseas data cards cost substantially more, and once that card cost is bound into the bill of materials, it lifts the unit price in the markets you were trying to win.
What we changed to
An Apple-style tray, so the user and the channel choose the card. The original worry, that a user damages a card and the responsibility comes back to the manufacturer, is handled with a tamper-evident seal. Units shipped with a card carry the seal, and breaking it to remove the card voids warranty. Once the boundary of responsibility is written down, the anxiety that drove the original decision goes away.
What I actually got wrong
The root of this was assuming I understood the user and the channel better than they did. Soldering the decision shut looks like taking responsibility, and it traded away flexibility and cost control to buy a smaller problem.
On hardware, when a choice can be handed to the user or the channel, hand it over, and spend the engineering effort on making the consequences of a wrong choice survivable instead.
Related
Want to try the hardware behind these notes?
Six AI hardware platforms are in mass production. Tell us your market and channel and we send an evaluation sample.
Request a sample