In this article
Two battery systems can both claim 98 per cent efficiency and still be a long way apart on what they cost you to run. The datasheet number is a peak, measured at one operating point on one piece of the system. It is not wrong. It is just not the number that separates a good system from a poor one.
The number that does that is the System Performance Index, and the 2026 results are worth reading before you sign anything.
Whole-system losses measured by SPI
HTW Berlin has run an annual efficiency comparison of residential battery systems since 2018. The 2026 edition, published on 5 March, tested 12 systems across four categories: 5kW and 10kW, each split into AC-coupled and DC-coupled.
Rather than measure a component at its best moment, SPI simulates a full year of household operation and counts four kinds of loss:
- Sizing losses, where the inverter or battery is the wrong size for the load it is asked to carry
- Conversion losses, the DC to AC and AC to DC steps, which is the only category the datasheet figure covers
- Control losses, where the system responds slowly or badly to a change in load
- Standby losses, what the equipment draws while doing nothing
It resolves to one percentage. Higher is better, and because the same simulated household runs against every system, the numbers compare directly.
The 2026 results
In the 10kW class, the Fox ESS PQ-H3-Ultra-10.0 hybrid inverter paired with an EQ3300-5 battery returned an SPI of 97.0 per cent, the highest figure recorded in the nine years the test has run. RCT Power, Energy Depot, Fronius, and Kostal with BYD all cleared 95 per cent in the same class.
At the other end of the field, one system returned 89.3 per cent.
That 7.7-point gap is the whole point of the exercise. Both ends of that range sit behind datasheets quoting peak conversion efficiencies in the high nineties.
The 5kW results were tighter and lower, which is expected: a smaller inverter carries its fixed overheads across less throughput. SAX Power Home Plus led the AC-coupled group at 93.2 per cent. Among the hybrids, SMA's Sunny Boy Smart Energy 5.0 with a Home Storage 6.5 battery took 92.8 per cent, with Kostal and BYD a tenth of a point behind at 92.7.
The cost of standby power
Standby draw is the figure we would look at first, and the one least likely to appear on a quote.
Doing nothing.
Still drawing power.
Source: HTW Berlin · 5 March 2026 ↗. Two tested systems, not a complete market ranking.
Limits of the comparison
Two honest limitations, because a number this useful gets over-read.
The first is that HTW's simulated household is German. German load shapes, German irradiation, German tariff logic. Perth is a different problem: far more generation, a much heavier afternoon air-conditioning load, and an export regime that pays 2c for most of the day. A system optimised for a Berlin winter is not automatically the right system for a house in Baldivis with ducted refrigerated air running from two o'clock.
The direction of the ranking still holds. Conversion losses and standby draw are properties of the hardware and they do not change when the sun moves. What shifts is how much each category matters, and in WA the standby figure matters more than it does in the German model, because our systems spend more of the year with surplus generation and less time cycling hard.
The second limitation is participation. Twelve systems is a small field, and manufacturers submit voluntarily. A brand absent from the results has not failed the test. It has not sat it.
Comparing efficiency on quotes
If two systems are within a point of each other on SPI, the difference is not worth paying a premium for. You will not notice it. Spend the money on getting the sizing right instead, because sizing losses are the category most often blown by a system specified to a sales target rather than to the household's actual consumption.
If the gap is three points or more, take it seriously. That is a real annual cost, and it compounds across the warranty term.
And if you are being sold on a peak efficiency figure alone, ask what the system returns across a year. The two questions worth putting to any supplier are the standby draw in watts and whether the product has been through an independent whole-system test. Both have plain answers, and the answers sort the field faster than any brochure.
Full results are published by HTW Berlin at solar.htw-berlin.de.