What actually fails in a smart home, and when.

The short version

Smart home hardware does not mostly fail the way people fear it will. The common worry is that the electronics stop working after the warranty expires. What we actually see is that most returns cluster in a small number of specific, boring, predictable places. The components people worry about least are often the ones that go first.

The headline number

About five percent of what we have installed has come back to us with a fault. Nineteen units in twenty are still doing what they were fitted to do and have never needed us again. That is across twenty-nine homes. It is a small enough sample that we will say so rather than dress it up. It is what we have, and it is more than anybody else in this category publishes, which is nothing.

The more useful finding is what is wrong with the one in twenty. In most of the panels that reach the bench, the failure is in the power supply. Not the touch layer. Not the glass. Not the firmware, not the Wi-Fi, not the app, and not any of the things people actually worry about when they hesitate over a touch panel.

That is worth sitting with, because it inverts the usual fear. The anxiety about smart switches is that the clever part will stop being clever. That the software will rot, or the brand will abandon it, or the touch surface will stop reading a finger. What we see instead is the least clever component in the box giving out: the bit that turns mains into the low voltage everything else runs on. It has been made much the same way for forty years, it is one of the cheapest things inside the panel, and it is replaceable.

Which is the whole argument for a local bench. The dominant failure on the most expensive component we fit is a cheap part, and it can be changed in an afternoon. That only happens if somebody is willing to open the panel rather than courier it back and quote you for a new one.

Failures by component

The sections below are the categories worth covering. Each needs the same four things: what goes wrong, roughly how often, at roughly what age, and what it costs to fix.

Touch panels

The most visible component and the one customers ask about most, because a dead panel is a dead board on a wall in the middle of the house.

The dominant fault is the power supply, as above. What that means in practice is that a dead panel is usually not a dead panel. It is a working panel with a dead feed, and the repair is a part change rather than a rebuild. We do not refurbish panels and we do not replace touch glass; we replace the component that failed and send it back to the wall it came off.

Retrofit modules behind the plate

Hidden in the back box behind the plate, so a failure here presents as one circuit that stops responding while everything else carries on. These are the modules used in every smart switch install.

Curtain motors

Mechanical rather than electronic, which makes motorised curtain tracks a different kind of risk. Wear rather than sudden death.

Water tank controllers and sensors

Sitting on a Jaipur terrace, exposed to 45°C summers, monsoon, and dust, doing the least glamorous job in the house. Why the tank job matters here is set out in how Jaipur supply behaves.

Locks, cameras and door phones

The category where a failure has the highest consequence, because it can leave somebody outside their own house. It is also why we publish what it costs to replace a part rather than quoting on the day.

The failures that are not hardware

A meaningful share of what gets reported as a fault is not a broken component at all. A router gets replaced and the Wi-Fi password changes. A brand pushes a firmware update that changes how something behaves. Somebody removes an app while clearing space on a phone. A cloud account gets closed.

These matter because they are the failures a standards-based system survives and a proprietary one often does not. That is the practical argument for leading with Matter and Zigbee rather than a brand's own ecosystem.

What we would not buy

The section that makes this page worth reading, and the one that is uncomfortable to write.

What this means before you buy

Three things follow from all of the above, and they are the reason we design the way we do.

Every wall switch keeps working. A component failure should cost you the smart behaviour, not the light. This is why we will not sell a house where one tablet on one wall is the only way to turn something on.

Standards first, proprietary named as proprietary. Matter and Zigbee wherever a good product exists at a sane price, so a dead brand costs you one device rather than the house. Where we specify something proprietary, it says so on your design sheet along with what happens if that company disappears.

Repairable locally beats replaceable remotely. A component we can fix on a bench in Jaipur in days is worth more to you than a marginally cheaper one. The cheaper one has to be couriered to another state. How the service bench works