Watch the unit's cooling performance data, because refrigerant loss has a signature: the temperature drop across the cooling coil shrinks, cooling runtimes stretch, and stages work longer to hold the same setpoint. Monitoring that reads the data your thermostats or energy management system already produce catches that drift days to weeks before anyone in the building feels it, no gauges and no site visit required.

What does refrigerant loss look like in the data?

Low charge does not announce itself; it erodes performance in a pattern that minute-level data makes visible:

Symptom What it looks like in monitoring data
Less cooling per run The temperature drop across the cooling coil shrinks week over week
The unit works harder Cooling runtimes stretch for the same outdoor conditions
Stages compensate Second stages engage earlier and run longer to do the same job
Comfort holds, then breaks Setpoint is maintained early on, then slips on the hottest afternoons, which is usually when the first complaint arrives

The critical detail is "for the same outdoor conditions." A unit running longer during a heat wave is normal; a unit running longer at the same outdoor temperature it handled easily a month ago is not. That is why every comparison has to be weather-normalized against the unit's own baseline, which is exactly how the detection methodology works.

Why don't thermostats catch it?

Thermostats alarm on comfort failure: the space is too warm, now. Refrigerant loss is gradual, and the equipment compensates by simply running more, so the thermostat sees a unit that is still hitting setpoint and stays quiet. By the time the space actually overheats, the slow leak has become an emergency service call at 2 to 5 times the planned cost, and a compressor that has been running hot and long on low charge for weeks, which is how a hundred-dollar leak repair cascades past $5,000.

How early can refrigerant loss be caught?

Gradual leaks typically give days to weeks of warning in the data before comfort fails. In one technician-confirmed case on a monitored quick service location, the platform flagged the degradation nine days before the repair visit; the full case study is here. Across all technician-confirmed repairs on the monitored fleet, not just refrigerant cases, 87% were flagged before the technician arrived, with a median of 6 days advance warning. Six days is the difference between scheduling a planned visit at standard rates and paying after-hours emergency pricing while a dining room heats up.

What do you need to detect it remotely?

An existing data source and software that knows what to look for. If your buildings have an energy management system or connected thermostats, the temperatures, runtimes, and staging data needed to spot refrigerant loss are already being produced, and a software-only monitoring layer reads them with nothing installed, no refrigerant line sensors and no gauges. The software-only vs hardware-bundled comparison covers when that approach fits and when it does not.

Frequently asked questions

Can monitoring tell me exactly how much refrigerant is lost?

No, and be skeptical of anything that claims otherwise from thermostat data. Monitoring shows that cooling performance is degrading in the pattern refrigerant loss produces and how fast it is getting worse; the technician confirms the cause and quantifies the charge. The value is the early flag, so that visit happens on your schedule instead of the failure's.

Could the same data pattern be a dirty coil or a failing part instead?

Sometimes, because several problems erode cooling performance. Different causes leave different signatures, and the platform reads them, but the honest answer is that a technician confirms the specific cause on site. Either way the unit needed attention, and catching the degradation early is the point regardless of which repair it turns out to be.

Will hot weather trigger false alarms?

Not by itself. Every unit is compared against its own weather-normalized baseline, so working hard in a heat wave reads as normal, while working harder than the unit's own history at the same conditions is what raises a flag. Problems also must persist across nights to escalate, so one strange evening alerts no one.

What does this cost?

Elite Energy Management runs $5 to $12 per rooftop unit per month on standard tiers with no hardware and no installation, full pricing here. One leak caught before compressor damage covers years of monitoring for that unit.

If your locations already have an energy management system or connected thermostats, a free 30-day pilot will show you what every rooftop unit's cooling performance actually looks like, first health scores within days, no hardware, no commitment.

Related reading: the refrigerant leak case study, how detection and validation work, and a plain-English guide to fault detection and diagnostics.