The verdicts at a glance
| Claim | Verdict | What the evidence says |
|---|---|---|
| Heat pumps do not work when it is cold. | Myth | Efficiency falls as it gets colder, but the units keep delivering more heat than the electricity they use. In the Electrification of Heat trial the median air source COP was 2.37 at −3°C and 2.3 at −5°C outside. |
| Heat pumps do not work in old houses. | Myth | The 742-home trial found no significant difference in air source performance by house type or age. The Catapult’s earlier findings said there is no property type or architectural era that is unsuitable for a heat pump. Old homes face more technical hurdles, though. |
| You need underfloor heating. | Myth | Air-to-water systems heat water that can feed radiators, underfloor heating or both. Radiators often need to be larger because the water is cooler. |
| You must insulate the whole house first. | Myth for the grant; partly true for comfort | The grant’s insulation condition was removed on 8 May 2024 and a valid EPC is no longer required (relaxed from 28 April 2026; the installer must still evidence the heating being replaced). Better insulation still shrinks the unit you need. |
| They are too noisy. | Mostly myth, with a siting catch | In the trial, participants said noise was not an issue, but 5% of properties could not meet the planning noise criteria for siting. The England rule is 37.0 dB(A) at the neighbour’s assessment position. |
| They are always cheaper to run than gas. | Not always | At the 1 October to 31 December 2026 cap our model puts heat from a heat pump at SCOP 3 at about 8.8p per kWh against 8.9p from gas: roughly level, not a landslide. |
| They are never cheaper than gas. | Myth | Against oil (about 13.0p per kWh of heat) and electric heaters (26.3p) they are clearly cheaper. Against gas a good design or a lower average tariff can tip it. |
| The grant is paid to me after the work. | Myth | The installer applies, Ofgem emails the property owner for consent, and the grant is deducted from the quote and invoice upfront. |
“They do not work when it is cold”
The physics: an air source heat pump extracts heat from outdoor air, and there is usable heat in air well below freezing. What happens is that efficiency falls as it gets colder, because the temperature gap the machine has to bridge grows. The trial measured this. Across its air source heat pumps, the median 30-minute coefficient of performance peaked at 3.37 when it was 10°C outside and fell to 2.37 at −3°C and 2.3 at −5°C, which the report notes are close to common design conditions in Great Britain. Ground source systems were much less affected: the median COP at −5°C was 3.1, though the ground source sample was small.
So the verdict is that the claim is a myth, with a fair concession. A COP of 2.3 still means more than two units of heat for every unit of electricity, but the coldest days are the most expensive ones to heat. Installers size the unit against the design temperature for your area so that the system holds the house at temperature on those days; the heat-loss calculation and flow temperature are the parts to check on a quote.

For the full winter data, by outdoor temperature, see do heat pumps work in cold weather.
“They do not work in old houses”
The Electrification of Heat trial installed 742 heat pumps, including 163 of them in properties built before 1945 (22% of the total). It found no significant difference in air source performance by house type or age, which the report puts as: where a trained installer recommends a heat pump for a property, house type or age should not affect performance. The same project’s earlier published findings were that there is no property type or architectural era that is unsuitable for a heat pump (December 2021).
The concession is in the same report. Homes built before 1919 were more likely to have technical barriers to installation than other types, and mid-terraces more practical barriers, mostly outdoor space and noise siting. Not “does not work”; “harder to design and sometimes impossible to site”. We go through it in heat pumps for old houses.
“You need underfloor heating” and “all your radiators will have to go”
A heat pump heats water that can feed radiators, underfloor heating or both. Underfloor works well because it runs at low temperatures, but it is not required. What matters is flow temperature. The trial measured median COP of 3.5 when the heat pumps were making 35°C flow, 2.38 at 55°C and 2.02 above 65°C. So radiators that let the house be warm with cooler water give a cheaper-to-run system, and some may need replacing or enlarging for that reason. How many is a result of the heat-loss survey, not a rule; it is a line to look for in a quote. See sizing and radiators.
“They are too noisy”
In the trial, participants reported that their heat pump kept the home warm, was reliable and easy to use, and that noise was not an issue. But sited badly, a unit can bother a neighbour, which is why England has a rule. Under permitted development an air source unit must meet the MCS planning standard, which sets 37.0 dB(A) at the assessment position, usually a neighbour’s window. In the trial, 5% of properties could not meet the noise criteria even with abatement because the unit would sit too close to a neighbour, mostly mid-terraces, and those needed planning permission. We cover it in heat pump noise and planning permission.
Verdict: mostly a myth for the owner, with a real siting question in tight terraces and flats.
“Heat pumps are cheaper to run” and “they are never cheaper”
Both are wrong as stated. At the 1 October to 31 December 2026 Ofgem cap, our labelled model of 12,000 kWh of heat a year costs about £1,053 with a heat pump at SCOP 3, £1,063 with a gas boiler, £1,566 with oil and £3,158 with direct electric heating. So: clearly cheaper than oil and electric, roughly level with gas. The trial adds a caution: the median measured efficiency of air source systems was 2.78 to 2.93 depending on how the system boundary is drawn, and the measured SPF was below the designed SCOP for 91.9% of properties. The full working is on the running costs page and the tariffs page.
The grant myths
- “You need a good EPC or full insulation.” The insulation condition was removed on 8 May 2024, and a valid EPC is no longer required (relaxed from 28 April 2026; the installer provides the EPC reference if one exists, otherwise other evidence).
- “The grant is paid to me afterwards.” The installer applies, Ofgem emails the property owner for consent, and the grant is deducted from the quote and invoice upfront.
- “Ground source gets less.” Both air-to-water and ground source get £7,500; the £9,000 band applies where the home has no mains gas and the heating replaced is oil or LPG.
- “Air conditioning units cannot get the grant.” Out of date: air-to-air heat pumps in homes have been eligible for £2,500 since 28 April 2026. See eligibility.
Where the sceptics are right
- Performance varies a lot with design and installation. The trial’s own authors link part of the spread to inconsistent quality.
- Designed SCOP numbers are often optimistic compared with measured results.
- On a standard tariff, running costs are about level with gas, not far below it.
- Upfront cost after the grant is still real money for many households.
- Some homes, such as tight terraces, are hard to site a unit on.
None of this is a physics problem, and all of it is worth checking before you sign. That is the point of the quotes checklist and the worth-it verdict.
Two more doubts, answered
“My friend’s heat pump never got the house warm.”
That happens, and the trial’s own findings point to the cause: a wide spread in measured efficiency (the middle half of air source systems measured 2.55 to 3.05 on the whole-system measure) that the report ties in part to inconsistent design and installation quality. A heat pump that is undersized, or running at high flow temperature through undersized radiators, can disappoint.
The remedy is at the quote stage: a written heat-loss calculation, design flow temperature and emitter schedule. See the quotes checklist.
“If it were that good, why do some installs fail to hit their SCOP?”
Fair question. In the trial the measured SPFH2 was below the designed SCOP for 91.9% of properties, with the median 17.9% lower. That is a gap between calculation and reality, and a reason to treat a quoted SCOP as an upper bound, not a promise.
Commercial premises are treated differently in every nation, and the specialists at Heat Pumps For Businesses keep a separate rundown of business funding.