The question everyone asks
Walk into any heat pump forum, online community, or installer conversation and within minutes someone asks: which brand should I choose? Vaillant or Mitsubishi? Samsung or Daikin? Is it worth paying the premium for a Viessmann? The question is entirely reasonable — for a £10,000–20,000 purchase that will heat your home for 15–20 years, brand reputation feels like it should matter enormously.
The problem is that almost all the comparison content available is manufacturer-funded marketing, installer preference, or anecdotal forum opinion. HeatpumpMonitor.org is the only publicly available UK dataset with enough systems, enough brands, and enough measurement rigour to actually answer the question with data.
// About the dataset
639 heat pump systems uploading continuous monitoring data as of November 2025, including 448 with full MID-certified metering (billing-grade heat meters and electricity meters). The analysis below focuses on 170 air source systems at the H4 boundary with no summer cooling, achieving a dataset average SPF H4 of 3.86. Systems span the full range of UK property types, installer quality, and commissioning approaches. The dataset includes systems installed by some of the best heat pump engineers in the UK — meaning it captures what good performance looks like, not just average.
The short answer
System design, flow temperature, and commissioning matter far more.
The data is unambiguous on this point. When you look across the full dataset, the strongest single predictor of annual performance (R² = 0.55) is not the brand name on the outdoor unit — it is the weighted average flow temperature minus outside temperature. Physics, not marketing. The lower the flow temperature relative to outside conditions, the higher the SPF. Every manufacturer's unit obeys this rule because it's thermodynamic law, not engineering preference.
What the equations actually say
The HeatpumpMonitor.org team have published the regression equations directly from the dataset. These are worth knowing because they tell you exactly what a flow temperature decision is worth in pounds per year.
90% prediction interval: ± 0.55 | R² = 0.515
Reading the equation directly:
| Weighted avg flow temp | Typical SPF H4 | ±90% interval | Assessment |
|---|---|---|---|
| 30°C | 4.4 | ±0.55 | Excellent — well-optimised system |
| 35°C | 3.9 | ±0.55 | Good — well-tuned radiator system |
| 40°C | 3.4 | ±0.55 | Moderate — room for improvement |
| 45°C | 2.9 | ±0.55 | Poor — likely sub-optimal commissioning |
Every 5°C reduction in average flow temperature adds approximately 0.5 to the annual SPF. On a typical UK home using 15,000 kWh of heat per year, the difference between SPF 2.9 and SPF 4.4 is the difference between consuming 5,172 kWh of electricity and 3,409 kWh — a saving of 1,763 kWh annually, worth approximately £489 at 27.75p/kWh. That's every year, for the life of the system.
What brand actually explains: the Carnot percentage
Once you control for flow temperature and outside conditions, what's left? The HeatpumpMonitor.org team measure this using the practical efficiency factor — the percentage of the theoretical ideal (Carnot) COP each system actually achieves. This is where any brand-level quality difference would show up.
Practical_COP = % Carnot × Carnot_COP
// Industry assumption: ~50% of ideal. Reality: 45–58% depending on system.
The impact of this percentage is substantial even in a narrow range. At the same system temperatures — 35°C flow, 6°C outside — the difference between operating at 45% and 55% of ideal Carnot COP:
| % Carnot | Ideal COP | Practical COP | Annual electricity cost* |
|---|---|---|---|
| 40% | 8.4 | 3.35 | £1,118 |
| 45% | 8.4 | 3.77 | £993 |
| 50% | 8.4 | 4.19 | £894 |
| 55% | 8.4 | 4.61 | £812 |
* Based on 15,000 kWh annual heat demand at 27.75p/kWh electricity. Conditions: 35°C flow, 6°C outside.
A £306/year difference between 40% and 55% Carnot. That's real. The question is whether it's explained by brand — and the answer from the data is: only partially, and less than you'd expect.
Carnot percentage ranges by brand — what the data shows
The HeatpumpMonitor.org team have published the % Carnot ranges they observe for each major brand across their dataset. The results are striking for what they reveal about within-brand variation:
Every brand has a substantial spread within it. Mitsubishi's best-performing systems achieve 56% Carnot; its worst achieve 39% — a 17-point spread within a single manufacturer's products. The variation between the worst Mitsubishi system and the best Viessmann system is only slightly larger than the variation within Mitsubishi alone.
The clear implication: choosing Viessmann over Mitsubishi, or Vaillant over Samsung, does not guarantee you the upper end of the Carnot range. A poorly commissioned Vaillant will underperform a well-commissioned Mitsubishi. The brand badge is a starting condition, not an outcome guarantee.
What actually explains the within-brand variation
If not brand, what causes a Mitsubishi system to achieve 56% Carnot in one installation and 39% in another? The OpenEnergyMonitor team have investigated this carefully and identified several contributing factors:
| Factor | Impact | Fixable? |
|---|---|---|
| Primary pipework length before heat meter | High — can drop COP from 4 to 3 | At design stage |
| Sub-optimal weather compensation settings | High — drives unnecessarily high flow temps | ✓ Yes — at commissioning |
| Rapid cycling in mild weather | Moderate — never reaches efficient regime | ✓ Yes — settings/sizing |
| Heat meter inaccuracies (sensor placement) | Low-moderate — MPE up to 7% | Partially at install |
| Buffer tanks, low loss headers, PHEs | Variable — poor balancing can hurt | ✓ Yes — design stage |
| Compressor operating speed (high RPS vs low) | Significant — same unit, 3.3 vs 4.6 COP | ✓ Yes — sizing/settings |
| Refrigerant charge / manufacturing defects | Low but real — undercharged systems | Warranty/service |
The most striking item on this list is compressor operating speed. Analysis of the Vaillant performance tables published on HeatpumpMonitor.org shows the same 5 kW unit achieving a COP of 4.6 at 40 rps and only 3.3 at 110 rps — at identical system temperatures. A heat pump delivering most of its heat at high compressor speed (because it's oversized, or weather compensation is set too aggressively) will consistently underperform the same unit running at lower speed — regardless of brand.
The commissioning gap: the UK's biggest heat pump problem
The most important comparison in the dataset isn't between brands — it's between HeatpumpMonitor.org's average SPF of 3.86 and the Electrification of Heat (EoH) trial's average SPF of 2.81. Both datasets cover real UK homes. Both cover real heat pumps. The difference is 1.05 SPF points — enormous in financial terms.
// The EoH vs HeatpumpMonitor.org gap
A peer-reviewed study published in November 2025 (Energy and Buildings, Elsevier) draws on both datasets to explain the gap. Its conclusion: "high efficiency is closely linked to operation at low flow temperatures... the study highlights several key factors including suboptimal weather-compensation settings that drive unnecessarily high flow temperatures, frequent cycling on room temperature, and extended operation at less efficient compressor modulation levels."
The EoH trial covered 165 Vaillant Arotherm+ systems. The OpenEnergyMonitor team inspected every single one manually and categorised the weather compensation quality as Good, OK, or Bad. The result: a significant proportion had sub-optimal weather compensation — running at higher flow temperatures than necessary, reducing their SPF by approximately 0.5–1.0 points compared to equivalently-sized and installed systems on HeatpumpMonitor.org with properly tuned settings.
This is the Vaillant aroTHERM+ — one of the most popular heat pumps in the UK, installed under a major government programme. The problem wasn't the hardware. It was the commissioning.
On a typical UK home with 15,000 kWh annual heat demand, the difference between SPF 2.81 (EoH average) and SPF 3.86 (HeatpumpMonitor.org average) is approximately 1,400 kWh of electricity per year — worth about £388 at the standard rate. Over 15 years of ownership that's £5,820 in wasted electricity. From systems that are largely the same hardware as the high performers.
What good looks like: the case study systems
The dataset includes individually published case studies at the high end. Three examples from the documentation are instructive:
| System | Heat pump | SPF H4 | Coldest day flow temp | Design flow temp |
|---|---|---|---|---|
| System 68 | 10 kW Viessmann | 5.0 | 33°C avg (36°C max) | 40°C |
| System 278 | 10 kW Vaillant | 4.8 | 33°C avg (35.5°C max) | 43°C |
| System 53 | 5 kW Vaillant | 4.5 | 31.3°C avg (36°C max) | 35°C |
These high-performing systems have one thing in common: they all run well below their design flow temperature even on the coldest days. System 68 was designed for 40°C and ran at 33°C. System 278 was designed for 43°C and ran at 33°C. This is the practical efficiency factor in action — flow temperatures well below design indicate that real-world heat loss is lower than the survey calculation predicted, and that the weather compensation has been tuned to reflect actual demand.
The brand question: a proper answer
Having looked at all the data, here is the most honest answer the dataset can provide to "which brand should I buy?":
The brand ranges overlap heavily. A well-commissioned Mitsubishi easily outperforms a poorly commissioned Vaillant. A well-commissioned Samsung can match the mid-range Viessmann performance. The top of the Vaillant range is close to the top of the Viessmann range.
This does not mean brand is irrelevant. Some legitimate brand-level considerations remain:
| Consideration | What the data says |
|---|---|
| Installer familiarity | An installer who knows a specific brand's commissioning interface thoroughly is more likely to tune it optimally. This is arguably the most important brand-adjacent consideration. |
| Controls and app quality | Vaillant's myVaillant and integration with third-party optimisers is currently leading. Better controls make proper weather compensation more accessible. |
| Modulation range | Some models modulate down further than others in mild weather, reducing cycling. This affects the low-end of the Carnot range more than the high-end. |
| R290 refrigerant | Vaillant and some others have moved to natural refrigerant R290 (propane, GWP=3). R32 units (Samsung, Daikin, Mitsubishi) are still widely installed. Environmental consideration, not performance. |
| Installer network depth | Vaillant, Mitsubishi, and Daikin have the deepest UK installer networks. More installers means more competition and easier servicing long-term. |