Energy monitoring is where a lot of money gets spent on the wrong resolution. People buy fifteen metering plugs when one clamp on the incoming supply would have told them more, or fit an expensive whole-house monitor and still cannot answer which appliance is responsible.
The three tiers answer different questions. Knowing which question you have is the whole decision.
The three tiers
Whole house
What it is: a reading of everything the property consumes, from a smart meter interface or a current clamp on the incoming conductor.
What it answers: what am I spending, what is my standing load, when am I using it, and is anything unexpected running. It also enables load management and solar surplus diversion, both of which need the total.
What it cannot answer: which device. You see a 400 W step and have to work out what caused it.
Cost: low, and this is where everyone should start.
Circuit level
What it is: clamps on individual circuits inside the consumer unit — lighting, sockets by area, oven, shower, heat pump, EV charger.
What it answers: which part of the house, and crucially it covers hard-wired loads a plug cannot reach. For an electrified home this is the tier that matters, because the heat pump and the charger are the two largest consumers and neither has a plug.
What it cannot answer: which specific appliance on a shared socket circuit.
Cost: moderate, plus an electrician, because it means working inside a live consumer unit.
Appliance level
What it is: metering smart plugs, or in-line modules.
What it answers: exactly what one device does, including cycle detection and fault detection.
What it cannot answer: anything hard-wired, and it will not give you a household total unless you meter everything, which nobody does.
Cost: low each, high in quantity.
Which to buy, by situation
“I want to cut my bill and do not know where to start.” Whole house first, then rotate two metering plugs around the house a week at a time. That combination costs very little and answers most questions.
“I have a heat pump or an EV charger.” Whole house plus circuit clamps on those two. You cannot manage what you cannot measure, and both are large enough that a few per cent matters.
“I have solar and want to use more of it myself.” Whole house with import and export distinguished, plus generation. Diversion decisions need to know the instant you begin exporting.
“Something is using electricity and I cannot find it.” Whole house for the total, then circuit clamps to narrow it to an area, then a plug to confirm. Working top down is far faster than metering things at random.
“I want to know when the washing finishes.” A plug. That is all.
“I am on a dynamic tariff.” Whole house with a live feed, because the value comes from acting within the hour.
The specifications that matter
Does it work locally? The first question for anything. A monitor that only reports to a manufacturer's cloud gives you a dashboard rather than data, cannot drive automations reliably, and stops working when the service does.
Import and export separately. A simple clamp measures current without direction, so it cannot tell 2 kW coming in from 2 kW going out. If you have solar, you need a monitor that distinguishes them — either by reading the meter directly or by using a clamp that senses direction.
Update frequency. Once a second is useful for automation and appliance identification. Once every fifteen minutes is a report.
Accuracy at low loads. Most clamps and plugs are poor below a few watts. If you are hunting a 2 W standby load, no domestic device measures it reliably — measure a group of things instead and look at the difference.
Three-phase support, if you have a three-phase supply, which needs a monitor designed for it rather than three separate single-phase readings.
Data retention and export. Can you get your own history out, in a usable form? Trend analysis needs a year of data, and a service that keeps only a month is much less useful.
What to do with the data, which is the point
A monitor nobody looks at has saved nothing. Three things make it useful.
A baseline. Your standing load at 3am, and your daily total on a typical weekday. Everything is measured against these.
Comparisons rather than absolutes. This week against last week, this January against last January. Absolute figures are hard to interpret; changes are obvious.
Alerts on change, not level. A standing load 100 W above baseline means something is on. A daily total 30 per cent above normal means something is wrong. Both are more useful than any threshold.
Then a small number of automations that use it: shifting loads to cheap hours, diverting surplus, throttling a charger, and notifying on anomalies. Everything else is interest rather than saving, which is fine as long as you know which is which.
Where monitoring stops helping
Two honest limits.
Monitoring does not save energy. It tells you where energy goes. The saving comes from changing something, and the research on energy feedback suggests the effect fades over months unless it leads to a permanent change — a schedule, a setting, a replaced appliance, an insulated loft. Aim for permanent changes rather than continuous attention.
Appliance disaggregation is oversold. Systems claiming to identify every device from a single whole-house feed work reasonably for large distinctive loads and poorly for everything else. Treat the appliance breakdown in such products as an estimate, and use a plug when you need to know.
Gas, water and heat, which people forget
Electricity gets all the attention because it is the easiest to measure. In a gas-heated house it is frequently the smaller half of the bill.
Gas is worth measuring if heating is your largest cost, which in most European homes it is. Options are a smart meter feed where your country provides one, a pulse or optical reader on the meter, or in the absence of either, weekly manual readings written down — which is unglamorous and perfectly adequate for spotting a trend.
The useful comparison for gas is consumption against degree days — a measure of how cold the period was, so you compare a mild January with a harsh one fairly. Without that correction, a lower bill may just mean a warmer month. Several free services publish degree day figures by location.
Heat meters measure the heat actually delivered, in kilowatt-hours, by combining flow rate with the temperature difference between flow and return. On a heat pump this is what lets you calculate real efficiency — heat out divided by electricity in — rather than trusting a manufacturer’s figure. If you have a heat pump and want to know whether it is performing, this is the measurement that answers it, and it is the one most installations lack.
Water matters mainly for leak detection rather than cost, and the overnight test covers most of it.
What we would buy, in order
- A local whole-house feed — meter interface if your meter offers one, otherwise a direction-aware clamp.
- Two metering plugs to rotate around the house.
- Circuit clamps on heat pump, EV charger, immersion, oven and shower if you have them.
- A permanent plug on the freezer, washing machine and dishwasher for fault and cycle detection.
- A year of retained history, exportable.
- Three alerts: standing load above baseline, daily total well above normal, and expected consumption absent.
- One visible display of live cost for the household.
Items one and two together cost less than a set of eight plugs and will tell you considerably more. Start there and buy the rest only once you know what you are looking for.