There is a distinction that decides whether home automation saves you energy or merely saves you money, and almost no product page makes it. Some measures reduce how many kilowatt hours a house consumes. Others move the same consumption to a cheaper hour. Both are worth having. Only one of them is efficiency.
The short version
- Reducing load lowers both consumption and cost. Shifting load lowers cost only.
- Heating and cooling control is where the measurable savings are.
- Verified savings are consistently smaller than assumed savings, by a wide margin.
- Automation multiplies a good building envelope. It does not compensate for a poor one.
Reducing load and shifting load are not the same achievement
Reducing load means the house genuinely uses less energy: the boiler runs fewer hours, the heat pump modulates down, a room that nobody is in stops being heated to twenty-one degrees. Shifting load means the same energy is consumed at a different time, typically to catch an off-peak tariff or to soak up what your own solar array is producing at midday.
Shifting is valuable. It lowers bills, it reduces strain on the grid at peak, and with a time-of-use tariff it can be worth more than a modest efficiency gain. It is simply not the same claim, and conflating the two is how a system that moved your dishwasher to two in the morning gets sold as an environmental improvement. Ask which of the two a given feature performs before comparing it with anything else.
What the independently verified numbers actually say
The most useful reference point is the certification threshold rather than the advertising. Under the ENERGY STAR programme run by the US Environmental Protection Agency, a smart thermostat earns certification only if a year of real field data from hundreds of customer homes shows heating savings whose lower ninety-five percent confidence limit is at least eight percent, and cooling savings of at least ten percent, with additional floors of four and five percent on the weighted national average of twentieth percentiles. That is a deliberately conservative bar, and it is measured on runtime in occupied houses rather than in a laboratory.
Field evaluations tend to come in lower still. A billing analysis of residential smart thermostats covering 2019 to 2023, published by Energy Trust of Oregon in November 2025 and drawn from roughly 13,700 gas-heated homes, measured average savings of about 22 therms per home per year against the 39 therms the programme had assumed, a realisation rate close to fifty-six percent. Manufacturers’ own claims of fifteen to twenty-three percent sit well outside both of those ranges.
None of that argues against the technology. It argues for buying it with the right expectation, and for treating any percentage quoted without a sample size, a period and a heating fuel as marketing rather than evidence.
Where the savings actually sit, measure by measure
Heating and cooling dominate the energy use of most homes in temperate and cold climates, which is why control over them dominates the savings. Everything else is a smaller order of magnitude, whatever the packaging says.
| Measure | Reduces or shifts | Realistic effect |
|---|---|---|
| Learning or scheduled thermostat | Reduces | The main lever, in the high single digits to mid teens of heating and cooling energy |
| Room-by-room control on radiators or zones | Reduces | Meaningful in houses with rooms used at different times |
| Water heating scheduling | Both | Cuts standing losses and moves reheating to cheap hours |
| Smart lighting on top of LED | Reduces, marginally | Small, because the lamps are already efficient |
| Appliance scheduling and EV charging | Shifts | Cost savings on a time-of-use tariff, no reduction in kilowatt hours |
| Consumption monitoring | Neither, directly | Enables every other decision, which is why it comes first |
Why smart lighting saves less than people expect
Lighting was a large slice of the household bill when the lamps were incandescent. It is not any more, because regulation removed the inefficient stock from both major markets. In the United States, the Department of Energy’s backstop standard of 45 lumens per watt for general service lamps came into full enforcement in 2023, on manufacture and import in January and on retail and distribution in July. In the European Union, Regulation (EU) 2019/2020 applied from 1 September 2021 and withdrew the special-purpose exemption that decorative and carbon-filament lamps had been sold under.
Once every lamp in the house is an LED drawing a fraction of what it used to, switching one off automatically saves a fraction of a fraction. Smart lighting earns its place on comfort, scene control and security. Presenting it as an energy measure overstates a real but small effect.
Three mistakes that cost more than the equipment
Automating an envelope that leaks
A controller can decide when to heat. It cannot decide how fast the heat leaves. In a poorly insulated house with draughty openings, control simply meters out a loss that was always going to happen, and the percentage saving applies to a much larger base. Insulation, airtightness and glazing come first in any honest ranking, with automation as the multiplier afterwards.
Claiming a saving with no baseline
Consumption falls in a mild winter whatever you install. Any comparison worth trusting normalises for weather, usually through degree days, and compares like periods. Commercial energy management has formal methods for this, and the domestic version is simply the discipline of recording a full year before and after rather than comparing one January to another.
Letting comfort settings quietly undo the gain
The most common cause of underperformance is not a technical fault. It is a set-point that crept upward, a schedule overridden into permanent hold, or a room that now stays warm around the clock because it was easier than adjusting the app. Savings come from setbacks actually happening, which is why the systems that perform best are the ones people leave alone.
What to look for now, if you are buying
Two developments are worth knowing about. The first is interoperability at the energy layer. Matter 1.5, published by the Connectivity Standards Alliance on 20 November 2025, added an electricity tariff device type, support for time-varying tariffs, and electric vehicle charging with state of charge and bidirectional flow. That is the infrastructure that makes tariff-aware load shifting work across brands instead of within one app.
The second is measurement. Half-hourly or quarter-hourly data from a smart meter, matched against a controller that logs runtime, turns a vague impression into an auditable figure. Buy the visibility before the automation, because it is the only thing that will later tell you whether the automation worked. Tariff structures, incentives and metering rules differ substantially by country and by supplier, so check what applies where you live before assuming a payback period.
The full residential picture
Beyond heating control, the sequence in which measures are worth doing changes the payback considerably.
Read our complete guide to optimising home energy efficiency
The commercial equivalent of all this is a discipline in its own right, with metering classes and normalised baselines. We cover it in our explainer on how energy management systems work.
Published in 2025. Updated on 14 August 2026. Sources: ENERGY STAR smart thermostat key product criteria, US Environmental Protection Agency; Energy Trust of Oregon, Billing Analysis of Residential Smart Thermostats 2019-2023, published November 2025; US Department of Energy general service lamp standards, enforced 2023; Regulation (EU) 2019/2020, applicable from 1 September 2021; Connectivity Standards Alliance, Matter 1.5 release, 20 November 2025.
