Your Thermostat Could Be Costing You More Than You Think

Extreme weather due to climate change is putting increasing strain on our energy grid. Heatwaves in summer months can cause surges in air-conditioning that lead to outages, such as the 2025 heatwave that led to blackouts for 80,000 from Utah to Maine. In winter, storms can knock out generating capacity and cause spikes from heating increases at unexpected times, such as the 2021 ice storm in Texas that caused blackouts for some customers and astronomical bills for others.

To head off these freak events, power companies have had to dramatically increase supply.

“There are just a few hours a year that put a huge pressure on the grid,” says Saed Alizamir, associate professor of business administration in the Data Analytics & Decision Sciences (DADS) area at the University of Virginia Darden School of Business. “In order to respond to this kind of fluctuation, however, we have to maintain power plants that are both dirty and expensive.”

If behavioral patterns hold, Texas alone will require an additional 21,700 megawatts of generational capacity by 2050 to stay resilient, a rise of 20%–25% in grid operating costs.

Consumer heating and cooling from the residential sector are responsible for the lion’s share of pressure on the grid, says Alizamir—and could also hold the key to reducing that pressure.

“What are the things we can do to address those kinds of high fluctuations in energy consumption?” he asks.

To answer that question, Alizamir set out to understand what drives people’s energy use. “You’ve got to understand what’s going on and what drives people’s behavior,” he says.

His findings appear in “The Impact of Climate Change: An Empirical Analysis of Smart Thermostat Data,” recently published in the journal Management Science.

Using a unique trove of data showing hour-by-hour changes in consumer behavior, Alizamir and fellow researchers identified patterns that could lead to changes for the better. Alizamir wrote the paper with Michael Blair of Wilfred Laurier University in Toronto, as well as Shouqiang Wang of the University of Texas at Dallas. (Both Alizamir and Wang were Blair’s doctoral advisors at Yale University.)

Ordinarily, it’s difficult to get highly specific data on consumer energy usage. “You can look at monthly electricity bills, but they only give you aggregate data,” says Alizamir, who has long researched energy consumption and sustainability.

By contrast, the researchers were able to obtain highly specific information from smart thermostat company Ecobee, which allows customers to opt into a program called Donate Your Data (DYD). It shows when thousands of households in North America turn their thermostat on or off, change the temperature, or use pre-programmed settings.

“The data is extremely granular, showing changes at intervals of five minutes,” says Alizamir. “Because we also have data on outdoor temperature, we could also look at how that temperature affects people’s behavior.”

The Benefits and Perils of Automation

Alizamir and his colleagues looked at two types of behavior: long-term choices, including how consumers set the mode on their thermostat to either heating or cooling; as well as short-term decisions on how they set the temperature every day. In both cases, customers had the choice to automate functions, which affected energy usage in different ways.

Over the long term, they found customers who set the mode manually used heating in the winter and cooling in the summer. During the spring and fall, however, they tended to turn the thermostat off, saving energy.

By contrast, some customers used “auto mode,” which decides when to use heating and cooling modes, year-round. During the shoulder seasons of spring and fall, the thermostat is constantly changing back and forth between modes, using more energy.

“They essentially outsource that decision, so people who use auto are going to consume more,” says Alizamir.

When it comes to short-term decisions, automation had the opposite effect.

Conscientious consumers can set a program that automatically turns down temperature at night or while they are at work, saving energy overall. Some consumers, however, override the program with a “hold” function that increases or decreases the temperature, and then forget to set it back to the original schedule.

“For that reason, people who use ‘hold’ tend to consume more, because it leads to decision inertia,” Alizamir says.

In other words, automation in the long-term leads to using more energy, but in the short term it leads to using less energy.

“Understanding these behaviors are critical to changing patterns,” Alizamir says.

Re-Designing Thermostat Usage

Alizamir and his colleagues have shared these insights with the thermostat company, encouraging them to think differently about how they design their products.

“Not only is it good for the environment, but it’s also good for customers, because they save on electricity,” Alizamir says.

For long-term decisions, one possibility would be to remove the “auto mode” altogether when it comes to heating and cooling modes, avoiding the costly switching back and forth between modes in shoulder season.

For the short-term issues, notifications could be added to remind customers when they’ve left the thermostat on “hold” for too long—or the thermostat could just automatically revert back to the program after a certain length of time.

Another possibility would be to obviate the need for using “hold” by instituting a more gradual period of pre-heating or pre-cooling that would allow the temperature to gradually come up to the desired level.

“That way, when they arrive home, customers aren’t feeling too cold, so it doesn’t trigger them to over-react and put it on ‘hold’,” Alizamir says. 

Smart thermostats could also take advantage of “dynamic pricing” in some parts of the country, where electricity costs more during peak usage periods. By communicating with the grid, thermostats could heat or cool the home at times when energy is cheaper, saving money and reducing strain on the system.

Outside of the specific example of thermostats and the energy grid, Alizamir says, data collection programs like DYD can help companies better understand consumer behavior and be more responsive to their needs. “Many consumers assume customers are rational, and always make the best decision based on utility,” Alizamir says.

As with the case of using “auto mode” even though it costs more money, or putting the thermostat on “hold” and forgetting it, however, consumers don’t always act in their best interests.

Acquiring more granular data about how consumers use products and services can help provide valuable insights into how their actual behavior differs from the ideal.

“Customers show cognitive limitations that we cannot infer without dipping into the actual data,” he says. For example, they might react to price cues on a retail website in a way that can be tracked by analyzing online shopping patterns. “This kind of granular data helps us identify patterns that might not otherwise be apparent,” Alizmir concludes, “helping us better understand these decision-making paradigms.”

That, in turn, could help companies design better systems to benefit consumers, the company, and the environment.

Professor Saed Alizamir is co-author of “The Impact of Climate Change: An Empirical Analysis of Smart Thermostat Data” with Michael Blair of Wilfred Laurier University in Toronto and Shouqiang Wang of the University of Texas at Dallas, published in Management Science (2026).