Smart plugs, and what their energy numbers are worth

What the plug is actually measuring#

Inside a metering plug there are two measurements and one calculation.

• Current, usually through a shunt resistor or a small current transformer in series with the load.
• Voltage, if the design bothers. Better plugs sample the mains waveform directly. Cheaper ones assume a fixed 120 V.
• Power, computed by multiplying the two. Done properly, the chip samples voltage and current many times per cycle and averages the product, which gives real power and handles power factor correctly. Done cheaply, the plug multiplies measured current by an assumed voltage, which gives something closer to apparent power and overstates anything with a poor power factor: motors, older power supplies, fluorescent ballasts.

Cumulative energy is then integrated over time, either on the device or in the vendor's cloud. That is where a second class of error creeps in. A reboot, a firmware update or a lost connection can lose a chunk of the running total or double-count it, which is why a plug's lifetime kilowatt-hour figure often drifts away from reality even when its instantaneous watts look sensible.

How accurate is it, really#

Vendors of consumer plugs almost never publish an accuracy class, and there is no credible independent measurement of low-cost plug accuracy at low wattage that we can point you at. So the honest framing is structural rather than numerical:

• These are not revenue-grade meters. Nothing about them is calibrated or certified for billing.
• Accuracy is worst exactly where people care most: small standby loads of a few watts, where the sensing front end is closest to its noise floor.
• Sampling interval matters. A plug that reports every 10 or 30 seconds will miss a compressor's startup spike entirely, so the peak you never see is the peak that matters for sizing anything.
• Two plugs from the same brand can disagree, because there is no per-unit calibration step in a device that sells for a few dollars.

Use them comparatively. "This device draws roughly ten times more than that one" is a conclusion a smart plug can support. "This device used exactly 42.7 kWh last month" is not.

The plug's own standby draw#

Every metering plug is powered all the time. The radio, the microcontroller and, on many designs, the relay coil consume energy continuously, and the plug never includes its own consumption in the figure it reports. So a plug reporting 0.0 W is still costing you something.

Three things drive that figure, and none of them are usually published:

• The radio. A Wi-Fi plug keeps an association with your access point alive. A Thread or Zigbee plug runs a much lower-duty radio. That difference is one of the strongest practical arguments for Thread over Wi-Fi in a category you might buy twenty of.
• The relay type. A latching relay uses energy only at the moment it changes state. A conventional relay holds its coil energized the whole time the plug is on.
• The measurement circuit itself, which is always running.

The irony is worth stating plainly: putting a monitoring plug in front of a 1 W device to save that watt can be net negative. Monitor things that draw real power, and switch things you actually want switched.

What Matter exposes, and who shows it#

Matter has two relevant clusters: electrical power measurement and electrical energy measurement. Both arrived in Matter 1.3 (May 2024), and both can be attached to any device type, not just plugs.

That release date is the single most useful fact when shopping. The device itself must be Matter 1.3 or newer to report anything. Large numbers of plugs sold in 2022 and 2023 physically measure power perfectly well and have no cluster to report it over Matter, so the data appears in the vendor app and nowhere else. A "Matter smart plug with energy monitoring" is two claims, and the second one depends on the firmware's Matter version.

Then there is the controller half of the problem.

So an Apple household buying a metering plug for Apple Home is buying a switch. The measurement exists on the device and there is nowhere in the app for it to appear. Whether that changes in a future release is not something anyone has committed to.

Plugs documented to report energy over Matter include Eve Energy and the Aqara Power Plug H2 on Thread, IKEA GRILLPLATS on Thread, and the TP-Link Tapo P110M, Meross MSS315 and Ledvance Smart+ Plug on Wi-Fi. Behavior is firmware dependent, so the same model can differ between production runs. The broader energy picture, and why almost nothing in the category has shipped, is in Matter for energy.

When a plug is the wrong tool#

A smart plug is a small relay with contacts rated for a specific job. Several common loads are outside that job, and the failure mode is usually a relay that welds shut, which means a device you can no longer turn off.

• Motors. Compressors, pumps, power tools, window air conditioners and large fans draw an inrush current several times their running current at startup. The relay sees that surge on every switch-on. Contacts pit, then weld.
• Heaters and other high resistive loads. A 1,500 W space heater sits near the limit of a 15 A branch circuit and runs continuously for hours. Heater manuals routinely tell you to plug the appliance directly into a wall outlet for exactly this reason, and a smart plug is one more connection in a chain carrying near-maximum current for a long time.
• Anything that must not restart unexpectedly. Space heaters, irons, and any appliance with a heating element. A plug restores power after an outage on its own schedule, and the appliance may have no idea it was ever off.
• Anything that must not stop. Freezers, sump pumps, aquarium equipment, medical devices, network gear you rely on to fix the smart home. A welded-open relay or a failed firmware update becomes a real problem rather than an annoyance.
• Devices with a soft power button. A plug restores mains power, but a TV or console that needs a button press after power returns will simply sit there. The plug did its job and the appliance still did not turn on.

Outdoor use and ratings#

Outdoors is a different product category, not the same plug in a bag.

• Weather rating. Look for a published ingress rating on the plug itself. "Weather resistant" with no rating on the spec sheet is marketing.
• GFCI protection. Outdoor receptacles in US homes are required to be GFCI protected under current code. A smart plug does not provide that and must not be treated as a substitute.
• In-use covers. An outdoor receptacle with something plugged into it needs a cover that closes over the plug body. Many smart plugs are too bulky to fit under one, which quietly turns a code-compliant outlet into a non-compliant one.
• Temperature range. Relay and radio specifications degrade in cold. If the plug is for holiday lighting in a northern winter, check the operating range before December.
• Range. Wi-Fi at the far end of the yard is worse than it looks. This is one of the places Z-Wave Long Range is genuinely the better answer.

What plugs are genuinely good at#

• Cycle detection. Watch the power draw of a washer or dryer and fire a notification when it falls below a threshold for a few minutes. This is the single best use of a metering plug, and it works fine with imprecise data.
• Vampire load hunting. Move one plug around the house for a week, note what draws power doing nothing, and act on the big ones.
• Scheduling and seasonal control. Holiday lights, dehumidifiers, towel rails, trickle chargers.
• Making a dumb appliance observable so an automation can react to it, without touching the appliance itself.
• A cheap first Matter device. The on/off plug-in unit is one of the most widely certified device types there is, which makes it a low-risk way to test commissioning on a new platform before you commit to anything expensive.

How to choose#

• For switching only: any certified plug on Thread or Zigbee. Cheap, small, no cloud in the path if your controller is local.
• For energy data you will look at: a Matter 1.3 or newer plug, plus a platform that displays it. Home Assistant, SmartThings and Homey today.
• For energy data you will act on: skip plugs and meter the circuit.
• For outdoors: a rated outdoor unit, on a GFCI-protected receptacle, with the range problem solved before the weather problem.
• For a whole house: stop buying plugs at some point and use smart switches or in-wall relays instead. Twenty Wi-Fi plugs is twenty clients on the band your sensors are trying to use.