The COP number on the box isn't the COP you'll actually get

I had a client in the Dandenong Ranges email me a spec sheet a few months back, genuinely annoyed. His new ducted heat pump was rated at a COP of 4.8, and he was getting nowhere near that on the coldest nights of the year. He wasn't wrong to expect better. But he'd misread what the number actually promised him, and honestly, most people do.

The coefficient of performance, or COP, tells you how many units of heat a system delivers for every unit of electricity it consumes. A COP of 4 means four units of heat out for one unit of power in. It's a genuinely useful number, and it's the reason heat pumps beat resistive electric heating or gas by such a wide margin. The problem isn't the concept. The problem is that the number printed on a brochure is a single snapshot, taken under one specific set of lab conditions, and your loungeroom on a July night in Ballarat is not a lab.

What the advertised COP actually measures

Manufacturers test COP under standardised conditions set out in bodies like AS/NZS 3823 in Australia or the equivalent AHRI testing protocols in the US. Typically that's an outdoor temperature around 7°C and an indoor temperature around 20°C, sometimes with results also published at a milder rating point. It's a fair, repeatable benchmark, and it lets you compare one unit against another on equal footing. That's genuinely valuable.

What it doesn't tell you is what happens at minus 2°C, or on a still, damp, 3°C morning with the compressor working against a frosted-up outdoor coil. As the temperature differential between outside air and your target indoor temperature grows, the refrigeration cycle has to work harder to move the same amount of heat. COP drops. Not because anything is broken, but because that's how a heat pump's physics works. I wrote a bit about the mechanics of this in how heat pumps actually work, and it's worth understanding the refrigeration cycle before you take any spec sheet number at face value.

Why the gap is bigger than people expect

On a mild Melbourne autumn evening, a well-sized heat pump might genuinely deliver close to its rated COP, or even beat it slightly. On a sharp Canberra frost, or a Minnesota cold snap, that same unit could be running at half its rated efficiency, sometimes less. It's still miles better than a resistive electric heater, which never exceeds a COP of 1. But if a homeowner budgeted their winter running costs off the brochure number, the bill shock is real.

This is exactly why the US rating system, HSPF (Heating Seasonal Performance Factor), exists as a seasonal average rather than a single-point figure. It's a better real-world indicator than a lab COP alone, because it accounts for a spread of outdoor temperatures across a heating season rather than one cherry-picked test point. Australia's rating system is moving in a similar direction with seasonal performance factors, but a lot of retail spec sheets still lead with the single peak-condition COP because it's the biggest, most flattering number. I'd argue that's a bit misleading, even if it's not technically false. Manufacturers aren't lying, they're just showing you their best day.

The performance curve is the number that actually matters

If you want to know what a unit will really do in your climate, ask the installer or check the technical datasheet for the full performance curve, not the single rated COP. A proper datasheet will show capacity and COP across a range of outdoor temperatures, often down to minus 15°C or lower for cold-climate units. That curve tells you the story a single number can't.

Two units can share an identical headline COP of 4.5 and behave completely differently at minus 5°C. One might hold 80% of its rated capacity, the other might drop to 50% and lean on electric resistance backup strip heating to make up the shortfall, which quietly tanks your real seasonal efficiency. This is precisely what the "cold-climate rated" label is meant to signal, though as I've covered before, that label doesn't guarantee what you'd assume. It's worth checking the actual curve rather than trusting the sticker.

Where oversizing and load calcs quietly make this worse

A unit that's oversized for the space often looks like it's coping fine in cold weather, because it's got so much spare capacity it never really strains. But it short-cycles constantly, which tanks efficiency and comfort in a completely different way. I've gone through this in more detail in why oversizing quietly ruins comfort and running costs, and the underlying cause usually traces back to a skipped or fudged load calculation, which I covered in Manual J or gross feeling. Get the sizing wrong and the COP conversation becomes almost academic, because you're fighting a bigger problem.

Similarly, the design temperature you or your installer chooses for the sizing exercise matters enormously here. If your installer sizes off a mild average winter day instead of the genuine cold-snap design temperature for your postcode, the unit will be undersized for the nights that actually matter, and no amount of COP will save you. I go through that mistake specifically in the design temperature mistake.

What actually happens at your real running cost

Here's the practical version. Take the rated COP with a grain of salt and instead look at three things: the manufacturer's full capacity-and-COP curve against outdoor temperature, your region's actual winter design temperature (not the seasonal average), and whether the unit has a backup resistance element that kicks in below a certain point, because that backup heat runs at COP 1 and will spike your bill on the coldest nights regardless of how good the compressor is.

For a rough real-world running cost estimate, I'd rather work backwards from your last few winter electricity bills on a similar or older system than trust a manufacturer's brochure math. I've laid out that comparison process against gas in heat pump vs gas heating, and the same logic applies whether you're comparing to gas or just trying to sanity-check a quote. The US Department of Energy's ENERGY STAR programme publishes decent guidance on how HSPF and SEER2 ratings should be read in context, and it's worth a look if you're comparing US-market units, because the labelling conventions differ from the Australian AS/NZS approach in ways that trip people up when they're reading imported spec sheets.

Noise, frost cycles and the number nobody puts on the box

One thing that never makes it onto a spec sheet at all is defrost cycling. In genuinely cold, humid conditions, the outdoor coil frosts up and the unit periodically reverses to melt it off, which temporarily drops delivered heat and can make the unit sound like it's working harder, because it is. That's normal operation, not a fault, and I've covered the noise side of that separately in why your heat pump sounds louder in winter. But it's another reason the lab-condition COP and your lived experience on a frosty Tuesday morning are two different things.

None of this means COP is a useless number, or that you should ignore it when comparing units. A unit rated 4.8 is still, all else equal, better than one rated 3.2. Just don't multiply that number by your electricity price and assume that's your winter bill. Ask for the performance curve, check the design temperature for your postcode against the AHRI or AS/NZS test points, and if you're shopping cold-climate options, look at products like the Zuba Central or Daikin Premium Inverter Ducted range, both of which publish extended low-temperature performance data rather than just the single headline figure.

Getting a quote that actually reflects your climate

If your installer's quote only references the single rated COP and doesn't mention your region's design temperature or a capacity curve, that's a fair sign the load calc was rushed. A properly vetted installer should be able to walk you through both. I've written a checklist for this exact conversation in how to vet a heat pump installer, and it's the single best hour you can spend before signing anything.

My mildly unpopular take: I think the industry should be legally required to publish the full performance curve on the front page of every spec sheet, not buried in a technical PDF three clicks deep. Until that happens, the burden's on you to ask for it.