Don't buy a promise. Ask for the numbers.

Researchers at the University of Nebraska–Lincoln measured 341 newly installed central air conditioners and heat pumps in 2022 and 2023. Their June 2026 report for the U.S. Department of Energy found 23% of systems seriously under- or overcharged, meaning at least 20% off the correct refrigerant level. Evaporator airflow was low in every city they visited. In lab tests, combined faults cut capacity by up to 42% and efficiency by up to 39% on a standard-efficiency system.^{Source 1} A companion dissertation estimated that faults in systems installed in 2023 add about 2.5 billion kilowatt-hours of electricity use a year nationally, about 9% above the same systems without faults.^{Source 2} The sample was not drawn to represent every home or any one city.

A 2026 study presented at the ACEEE Summer Study tested 83 recently installed systems in Arizona, California and Florida. The charge test was whether superheat or subcooling was within ±3°F of the manufacturer's target. 92% of 25 systems installed through quality-installation programs met it, which require recorded charge checks, training and follow-up inspections, compared with 55% of 58 systems installed by regular practice.^{Source 3} Most incorrectly charged systems were undercharged. When researchers offered the regular-practice contractors $400 for a visit to correct the charge, slightly more than half did not agree. The groups were not randomly assigned, the program in the study was run by co-authors of the paper, and the Arizona program homes were in the Tucson Electric Power area.

In a study published by ASHRAE in 1997, researchers tested 28 air conditioners in 22 new Phoenix homes. They recovered and weighed the refrigerant in 27: only 5 were within 5% of the correct charge, 21 were undercharged and 1 was overcharged. Airflow across the indoor coil averaged 14% below specification. In 23 of the 28 systems, all of the supply ductwork ran through the attic.^{Source 4}

In a 2021 laboratory study of a 3-ton R-410A system with a fixed metering device, a 20% undercharge cut cooling capacity 16 to 18%, and a 30% undercharge cut it 27 to 29%, compared with the same system correctly charged.^{Source 8} On a 3-ton system, a loss that size is close to a full ton of cooling.

In a 1996 Texas A&M laboratory study that simulated return-duct leakage, a 7.7% leak of 140°F attic air at 16.8% relative humidity cut effective capacity and efficiency by about 28%.^{Source 6} Where return ducts run through a hot attic, a leak pulls that air straight into the system.

In Florida laboratory-home experiments run in 2021 and 2022 and reported at the 2026 ACEEE Summer Study, researchers measured cooling-energy increases from about 4% for a single fault to 41% for a combination of faults.^{Source 3} Faults interact, sometimes amplifying each other and sometimes offsetting, which is why each one should be measured, not assumed.