Hardware

A 12.99 euro plug that costs about 2.50 euros a year to keep plugged in

Shelly's Plug M Gen3 does plenty of real things, saving energy is not automatically one of them.

Manish Singh/July 29, 2026/5 min read

In 2015 the NRDC metered homes in northern California and found that always-on load from devices nobody was actively using averaged around 23 percent of household electricity, which they scaled to roughly 64 billion kWh and 8 billion dollars a year nationally. That study is the foundation under almost every smart plug advertisement written since, including the line that made me save this one(Control smarter. Save energy. With Shelly Plug M Gen3). I want to walk the dates on this device, because the arithmetic that decides whether it saves you anything is the one number nobody puts on the box.

2015

The 23 percent figure is a household total, and it includes the router, the alarm panel, the set-top box, the well pump, the things you cannot cut at the socket without breaking something. More conservative numbers exist and they are worth holding next to it: a 2025 review in Energies puts US-derived residential standby at 5 to 10 percent, the older IEA figure was 188 TWh a year or about 2.2 percent of OECD electricity, Spain's IDAE says around 7 percent of a household bill. The UK's Energy Saving Trust gives a consumer-scale number instead, around 45 pounds a year in Great Britain from switching appliances off standby, and it explicitly warns that some PVRs and satellite boxes should stay powered. That is the honest range. Somewhere between a rounding error and a fifth of your bill, depending on what you own and how old it is.

October 2024, Shelly

Matter certification for Shelly gear was announced, and at CES in January 2025 the company introduced Gen4 with Zigbee and Matter built in from the start. The Plug M Gen3 sits below that line. Its Matter support arrived by firmware rather than by design, which matters when you are deciding whether to buy the cheap thing now or the multiprotocol thing later. Gen4 plugs in the US line already carry Zigbee. If you are building a house's worth of sockets rather than one, the Gen3 price is buying you a platform that is one generation behind on radios.

Spring 2025

Matter firmware, 1.6.0, reached most Gen3 models after a beta period, with the Plug S Gen3 having been the only plug shipping with Matter out of the box. What you get over Matter is basic: on, off, simple state. Energy history, scripting, virtual components, all of that stays inside Shelly's own stack or the Home Assistant integration. Shelly's own plug comparison page carries the sharp edge, scripting on some models may require Matter to be disabled, so the two headline features can be mutually exclusive on the same box.

That trade sits exactly where the money is for me. I live in Tallinn on a Nord Pool spot tariff, and the loads worth timing here are the water heater and the dishwasher, not the television's standby draw. The switching logic for that is a script, jisotalo's shelly-porssisahko does Nord Pool spot on-device, clever-PV covers Tibber, Rabot and EPEX-spot, the Spotmarket-Switcher project handles the Victron end. EU spot pricing is moving to 15-minute intervals, which makes on-device scheduling more valuable, not less. If enabling Matter takes the script away on this SKU, the Matter logo is the feature I would drop.

9th May, 2025

Regulation (EU) 2023/826 came into force, replacing 1275/2008 and 107/2009, with 0.5 W standby and 0.3 W off-mode limits arriving in 2027 and an estimated 32.5 TWh a year of savings by 2030. Read that against the plug. The appliance you were going to police is being regulated down to half a watt of idle, and the plug you would police it with draws roughly the same order of magnitude on its own. Vampire load as a category is being legislated out of new equipment, slowly, and what remains is old equipment. Measure the old thing, do not blanket-plug the new thing.

14th November, 2025

The user and safety guide for S3PL-30116EU is dated here, version 1.0. The body is 46 by 46 by 85 mm and 73 g, the switching rating is 240 V, 13 A, 3000 W resistive, and the circuit must be protected by a B or C characteristic breaker rated at most 16 A, minimum 6 kA, energy-limiting class 3. Wi-Fi is 2.4 GHz only, 802.11 b/g/n, and the radio declaration under 2014/53/EU is filed by Shelly Europe Ltd. There is an internal temperature sensor and overheating protection.

The restrictions are the part worth quoting to anyone buying one. Indoor use only, do not plug one into another, do not cover it, and do not plug direct-plug-in equipment into it, which the guide spells out as chargers, power adapters and plug-in transformers. That last one deletes the picture most people have in their heads when they buy a metering socket, which is the phone charger left in the wall. The manual is a stronger citation than any of the blog advice about heaters and portable ACs, and it already rules out half the demo use cases.

H2 2025, Eurostat

Household electricity in the EU averaged 0.2896 euro per kWh, of which 0.0837 was taxes and levies, with Ireland highest at 0.4042 and Germany at 0.3869. One watt held continuously for a year is 8.76 kWh, so at the EU average it costs about 2.54 euro annually. Shelly's own support page says device self-consumption is under 1 W for older generations and might be higher on the new ones depending on enabled features such as Bluetooth. Community power-model data for Shelly plugs lands around 0.6 W running and ~0.3 W in standby, which is 1.50 to 2.50 euro a year of electricity to measure electricity.

The plug lists from 12.99 euro. So the payback question is not whether the socket saves energy, it is whether the load behind it idles well above ~1 to 2 W for enough hours to clear both the plug's own draw and its purchase price. A 15 W idling amplifier, definitely. A modern TV built to the 2023/826 limits, no. This is the number Shelly does NOT print anywhere I could find for this specific model, and it is the only one that decides the purchase.

April 2026

Shelly Group SE entered the SDAX. The company is headquartered in Sofia, listed on the Bulgarian Stock Exchange and in the Prime Standard in Frankfurt as SLYG, with FY2025 revenue around 149.7 million euro and 9M-2025 revenue of 86.9 million, up 33.8 percent. The reply under the original post calling this Chinese dropshipping electronics is wrong about the company and partly right about the hardware, because Shelly's own investor report describes asset-light production through contract manufacturers. EU-domiciled brand, EU legal entity carrying the RED declaration, contract-built boxes. That is the accurate framing, and it is different from both the marketing and the sneer.

The genuine gap in the catalogue is sockets. There is no Type I plug SKU for Australia, New Zealand or mainland China, the Plug M Gen3 is a Schuko part, and the US line is a separate ETL-certified product at 15 A and 1800 W. For anyone on Type I the Shelly route is in-wall relays and PM modules fitted by a licensed electrician, not a plug at all.

July 2026, firmware 2.0.0

The stable rollout for Gen2 and newer landed, and the changelog reads like a list of things that were previously wrong: an OTA authentication bypass fix, brute-force protection, RFC 7616 nonce handling, pairing required for RPC over BLE, replay protection for encrypted BTHome advertisements, TLS certificate date validation, and an HTTP to HTTPS redirect when enhanced security is on. An OTA auth bypass being closed in 2026 tells you what the update path looked like before 2026. It also argues for buying Gen3 or Gen4 rather than a discounted Gen2, since some Gen2 models have stopped at 1.7.x for feature updates.

The eavesdropping reply on the thread is bait, and the real privacy point underneath it is better than the troll's version. Non-intrusive load monitoring research shows that household load curves reveal occupancy and individual appliance use, so appliance-level power data is behaviourally informative by construction. Shelly runs fully local, the web UI needs no internet, cloud can be turned off, MQTT is off by default. The catch worth knowing is that the app's historical energy data is stored in the cloud rather than on the device, per independent German testing, so if you want the history and not the exposure, the history has to live in your own database.

29th July, 2026

The Home Assistant question from the replies has a precise answer. Gen2 and newer non-battery Shelly devices can act as Bluetooth proxies forwarding advertisement data, with active or passive listening selectable in the options flow, but they do not proxy active GATT connections the way ESPHome does. So BLU buttons and BTHome sensors work, and anything that needs a live connection, a battery BMS, an Inkbird-style probe, a connection-based thermostat, does not. There is also an open HA core issue reporting Shelly BLE gateways interfering with Bluetooth networks, which is worth checking before you scatter twelve of them around a flat.

What is still unanswered from Shelly's own pages:

  • Measured idle self-consumption of the Plug M Gen3 specifically, in watts, not a family-wide "under 1 W" for older generations.
  • Whether enabling Matter disables scripting on this model.
  • Whether the plug ships with Matter enabled by default.
  • Whether it supports bidirectional, feed-in measurement, which a German review claims for Gen3 plugs generally.
  • Whether S3PL-30116EU appears on the CSA certified-products list in its own right, or only under family certification.

Independent testing has already reported deviations of up to around 11 percent against calibrated reference meters at low load on the Plug S Gen3, which is fine for a blog and not fine for billing, and low load is exactly where a standby hunt operates. Shelly Europe Ltd should put the measured idle draw of S3PL-30116EU next to the 3000 W line in the user and safety guide and reissue it as v1.1, with the Matter and scripting exclusivity stated on the same page. Until that sheet exists, the only way to know whether this plug saves you anything is to borrow a clamp meter and measure the load first.