Your Smart Thermostat Isn’t Here to Help You

But that doesn’t mean it’s useless.

An illustration of a smart thermostat with two red horns and little red eyes
Katie Martin / The Atlantic

Everything’s so smart now. Smartphones, smart speakers, smart lamps, smart plugs, smart doorbells, smart locks, smart thermostats. Smart things are smart not because they have smarts, but because they connect to the internet. Online connectivity allows them to be controlled, either locally or from afar—and in ways both visible and invisible.

The sales pitch for smart devices typically focuses on convenience. Rather than needing to fumble with a physical switch, you can turn a smart bulb on and off from bed (or from Bed Bath & Beyond), for example. A smartphone allows you to do work or doomscroll while you watch television and ignore your children. A smart thermostat allows you to tweak your home’s temperature from wherever. But why?

I found myself asking this question after two unrelated smart-thermostat events this week. First, my electric utility emailed yet another offer to save big on smart thermostats from companies such as Google and Ecobee. And second, I read a new paper from the National Bureau of Economic Research (NBER) challenging claims that devices like the ones sold by Google and Ecobee can conserve energy. The two seemed either entirely compatible (the utility wants me to conserve energy, and smart thermostats will help) or completely at odds (I want to pay for less energy, and smart thermostats don’t help). After looking into the matter, I’m less confused but more distressed: Smart heating and cooling is even more knotted up than I thought. Ultimately, your smart thermostat isn’t made to help you. It’s there to help others—for reasons that might or might not benefit you directly, or ever.

Despite predictions rehearsed in sales pitches for the things, smart thermostats may not reduce energy use, and they could even increase it. That’s the conclusion reached in the  NBER working paper, written by a group of economists at the University of Chicago and elsewhere. Based on 18 months of data from households using the devices, the authors describe a “null effect,” in which any efficiency gains the gadgets offer are offset by an unintended increased usage afforded by their convenience.

All thermostats are just sensors that read an ambient temperature, compare it with a desired one, and turn on or off an appliance in response. Normal thermostats hold a specific temp until altered. Programmable ones—which are electronic but not internet-enabled—change their target temperature over the course of the day or the week, saving energy (and money) when a house is empty or its occupants slumber.

Honeywell, whose devices were used in the study, didn’t respond to my request for comment. An Ecobee representative said that the company’s products “delivered persistent incremental savings” and suggested that the NBER paper, which relies in part on data collected from July 2012 to December 2013, draws conclusions from outdated information; a Google Nest spokesperson also called the paper an “outdated study of a single device.” But the paper’s authors dismissed those objections, telling me that the energy-savings forecasts are applicable to today’s devices.

People think—or thought, perhaps—that smart thermostats reduce consumption because they are supposed to do more than watch the clock and the mercury. A combination of sensors and internet-enabled data access allows them to monitor outside conditions, home occupancy, and HVAC-system performance in order to adjust furnaces and air handlers more automatically, and to transition between inside temperatures more effectively. That’s the automated-control part of the equation.

But smart thermostats increase something else: their users’ ability, and therefore propensity, to adjust them. This is supposed to be a good thing, because you can control them from anywhere in the world. Economists would reason that, given the chance to reduce costs and thus raise self-interest, such users would take advantage of the opportunity to reduce consumption accordingly.

But smart thermostats also dial up another form of self-interest: comfort. During a brief heat wave last week, temperatures where I live reached the upper 90s—hotter than normal this time of year. Sleepless, I wished it were cooler. But it could be! I just reached for the iPhone on my nightstand and lowered the thermostat by a couple of degrees. It’s a change that I wouldn’t have made had it required getting out of bed, a venue I enjoy for its soft flatness. But the smart thermostat meant I didn’t have to.

The cynical take on this outcome is that efficiency was never an earnest promise of smart thermostats, but merely an aspirational one. Instead, this position holds, the gadgets were designed to extract data from your home for aggregation, sale, or advertising, just like so many other things do in the ad-tech economy. This fear revealed itself only when Nest Labs, the company that made the reference product for the whole sector, sold to Google for $3.2 billion in 2014. Hold up, some Nesters thought, realizing: This thing is watching me.

They’re not wrong. Smart thermostats track lots of data, including temperature, humidity, ambient light, and motion. They use that information to run the thermostat, but it can also be used to detect or infer occupancy and activity. Once correlated with other data on the same network, including smart speakers and cameras, a smart-thermostat company can connect that data to specific users and their other behaviors, online and off.

Google is the usual suspect in such indictments, and the tech giant does indeed collect and use Nest devices in that way. (The company says that this is to “improve services.”) But it’s not alone. Even seemingly normal, boring home-thermostat makers such as Honeywell do the same sort of thing—and also reserve the right to share that data with third parties. In this account, the overall data economy becomes the beneficiary of the smart thermostat, which gets recast as a Trojan horse.

But hold on a minute. If that’s the case, why is my electric company encouraging me to buy a smart thermostat, and even subsidizing it? In light of their findings, the aforementioned economists think those subsidies “would be better spent on more effective interventions.” (Michael Price, an economics professor at the University of Alabama and one of the authors of the NBER working paper, told me that insulation or energy-efficient appliances might be a better target for subsidies to reduce actual energy usage.) Maybe so. But the utilities don’t encourage adoption of smart thermostats to produce individual household energy or cost savings. They do so to encourage residences to allow the utility to control their air-conditioning.

That sounds like it could be scary—it is, after all, giving an outside entity permission to control the temperature of your house—but there may be reason to embrace the practice. “Peak-load reduction,” as it’s called, is meant to reduce strain and increase efficiency on the overall electrical grid during the most intense periods of energy use. Doing so can help make electricity generation cleaner and avert blackouts and brownouts, which occur when the grid can’t meet electrical demand. This is an increasing concern for safety as much as convenience; people rely on power for medical equipment, but also to mitigate dangerous summer heat waves all across the country, made worse by climate change. To wit, faced with a major heat wave and record-high electricity demand earlier this month, California Governor Gavin Newsom asked residents to turn their thermostats to 78 degrees or higher in the evenings. Heating and cooling use more energy than anything else in the average American home, so modulating your AC can have a substantial impact.

Price told me that he and his colleagues didn’t investigate the benefits a smart thermostat could offer when controlled centrally by a utility, but he speculated that such control could help the devices operate “closer to what is assumed by the engineering models and would thus lead to greater savings than [those] observed in our study.” Ecobee also cited these programs as a benefit of its products, saying that 50,000 Ecobee customers voluntarily participated in peak-load reduction during the California heat wave this month. When carried out across a community, as peak-load programs aspire to be, reduced usage could be more significant, even if individual cost is not.

Peak-load reduction doesn’t require smart thermostats. Off-peak rate plans, along with the energy-source monitoring tools many utilities now provide, can also encourage customers to use less power at key times. But when utility customers willingly adopt smart thermostats and allow their electric companies to control them from afar, individual human preferences are taken out of the equation. That makes the community—or at least the electric grid—the beneficiary of smart-thermostat operations.

In each of these scenarios, the smart-thermostat user—that’s you and me—doesn’t play a leading role in the device’s operation. Instead, it’s data aggregators, or utilities, or municipalities, or the manufacturers who sell us these devices on incomplete, if not entirely false, promises. That’s a pretty strange situation to be put in by a box on the wall that controls your HVAC.

Then again, you can change the temperature from your bed. And tossing and turning over the knotted politics of the technology and energy sectors sure seems less appealing than being lulled into a cool slumber instead.