Phantom load audit: mapping every watt my house wastes at 3am

At 3:17am on August 14, 2026, my house drew 127 watts with every occupant asleep and every appliance supposedly off. That baseline—measured at the meter, verified device by device—translates to £186 annually at the current October 2024 price cap of 30.11p/kWh. This is not an estimate. This is what a £45 clamp meter and four hours of insomnia revealed.

The methodology: how to count watts at 3am without losing your mind

I used a TP-Link Tapo P110 smart plug for plug-in devices and a Fluke 376 FC clamp meter for hardwired circuits, logging readings at 3am to avoid solar inverter noise and variable loads like fridges cycling. Each device was isolated where possible, or its circuit traced through the consumer unit. Measurements were recorded in 30-second averages to catch pulsing loads. The full audit took three nights: one for mapping, one for measurement, one for verification. Raw data handling follows our standard practice—no cloud logging, local CSV only.

The big consumers: networking and entertainment

My router and mesh WiFi nodes consumed 18 watts combined—more than my refrigerator's compressor when idle. The television, ostensibly off, drew 12 watts in standby with quick-start enabled; disabling this dropped it to 0.7 watts but added 45 seconds to boot time. The soundbar added 8 watts regardless of input status. These three entertainment devices alone accounted for 20 watts, or £52.64 yearly. This is the penalty of convenience features marketed as "instant on."

Kitchen phantoms: the fridge that never sleeps

The refrigerator compressor cycled on for 8 minutes every 32 minutes at 3am, drawing 85 watts when running, averaging 21 watts across the hour. But the real surprise was the induction hob: 4.2 watts displaying a clock nobody uses. The dishwasher, finished hours earlier, held 2.1 watts for its moisture sensor and display. Combined kitchen standby—excluding active cooling—totaled 11.3 watts. Over a year, the hob clock alone costs £11.08. Insulation savings get attention, but these small drains compound silently.

The forgotten drawers: chargers and power bricks

I found 11 USB chargers and laptop bricks plugged in across four rooms. Individually they ranged from 0.3 watts (modern GaN charger, no load) to 5.8 watts (aging Dell laptop brick, battery full but connected). Total: 23.7 watts. Three were completely orphaned—devices long discarded, bricks still warm. The worst offender was a 2019 wireless phone charger: 3.4 watts continuously, delivering nothing. Collectively these drawers of forgotten electronics cost £62.34 annually, more than my entire lighting bill.

Heating and hot water: the controls that control nothing

The boiler's digital thermostat and circulating pump electronics drew 7 watts even with heating disabled for summer. The hot water cylinder's smart thermostat added 2.1 watts with its WiFi module pulsing every 90 seconds. Two motorized thermostatic radiator valves, installed for a zone system, consumed 0.8 watts each maintaining position memory. These heating-system phantoms totaled 12.7 watts—£33.21 yearly—during a month when no space heating was required. The infrastructure of comfort costs even when comfort is off.

Security and monitoring: necessary but not free

The doorbell camera, two indoor security cameras, and the smart door lock consumed 8.4 watts combined. The cameras were the bulk: 3.2 watts each, recording continuously to local storage. The doorbell's infrared LEDs and WiFi module added 2.1 watts. These feel essential, yet they represent £22.11 yearly in pure standby. I weighed replacing one indoor camera with a PIR-triggered model (0.1 watts idle, 6 watts active for 30 seconds per trigger) but calculated break-even at 4.7 years given the £89 hardware cost. The time-of-use penalty doesn't apply here—these loads are constant across all hours.

The measurement gaps: what I couldn't isolate

My consumer unit lacks per-circuit monitoring, so several loads remain estimates. The smoke alarm network (mains-wired, battery-backed) likely draws 2-3 watts but I cannot measure without disconnection, which violates rental terms. The loft's solar inverter consumes 8-15 watts at night for electronics and anti-islanding protection, but this is fused separately and I lacked safe access. These gaps total perhaps 15 watts unaccounted—£39 yearly in educated guesswork. Professional monitoring with a Wiser or similar would close this; payback is 2.1 years at current prices.

Intervention options: what actually reduces the baseline

Smart plugs with scheduling eliminated 11.4 watts from entertainment devices: £30 yearly for £44 hardware, payback in 17 months. Physically unplugging the orphaned chargers saved 9.2 watts immediately—cost zero, effort thirty seconds. Disabling quick-start on the television saved 11.3 watts but introduced latency I notice daily; this was reversed after two weeks. The boiler's summer mode reduced its draw to 3.2 watts, saving £9.96 yearly for a settings menu change. Combined actionable savings: 24.6 watts, £64.71 yearly, with £44 spent and no comfort lost.

The honest bottom line

My 127-watt baseline sits near the UK average of 100-150 watts for similar-sized homes, based on published monitoring studies. Elimination is impossible: refrigeration, security, and networking serve real functions. But the £186 in addressable phantom loads—devices pretending to be off—represents 13% of my annual electricity spend. The audit itself, four hours across three nights, has already paid for its time if I maintain the changes. The value is diagnostic: knowing which watts matter, which don't, and which trade-offs you're actually making.