The honest answer, before the analysis, is that we kept arriving at maybe. A small array generates the wrong shape of energy for a small house. The peaks are short, the dips are long, and the bits that match your evening load are exactly the bits you do not have. A battery promises to flatten this out. The brochure says it pays for itself in seven years.
So we measured. The roof was 4.1 kWp, south-facing, with a small chimney shading the bottom row from about 4pm in winter. The house was a 2010-build three-bed semi on the Surrey-Hampshire border with two adults and an EV. The tariff for the test year was Octopus Intelligent Go. We pulled the half-hourly export and import data from N3RGY directly, sized a 5 kWh battery against it offline, and asked four versions of the same question.
The narrow question
If you assume a battery is purely a self-consumption device, the maths is straightforward. Take every kWh you exported and ask whether a 5 kWh battery, charged from the panels and discharged in the evening, would have caught it. Over a year, the answer was a touch under a thousand kWh of export turned into self-consumption. At a delta of about 23p per kWh between export rate and import rate, that is around £230 a year.
A 5 kWh battery installed costs in the region of £3,500 to £4,500 in the UK at the moment, depending on chemistry and how integrated the install is. At £230 a year, that is fifteen to twenty years. The brochure number is a different number.
The wider question
Now widen it. A modern smart-tariff household does not just self-consume from solar. It cycles the battery from cheap overnight electricity, runs the house off it through the expensive daytime peak, and re-imports cheap electricity later if needed. On Intelligent Go that delta is closer to 18p per kWh on the right days. Over a year of disciplined cycling, that is several hundred pounds more, and now you are at a less alarming payback.
The complication is that this strategy stops being purely about your solar array. The array becomes the smaller of two energy sources for the battery. The thing earning you money is the tariff structure, not the panels. This is not necessarily a problem. It just means the case for the battery would still hold without the solar.
What we changed our mind about
Twice. The first time, we changed our mind from yes to no when we noticed that the modelled savings assumed a level of behavioural alignment with the tariff that real households rarely sustain. The second time, we changed our mind from no to maybe when we ran the same data against a household with no EV and a smaller daytime load, and found that the case there was actually stronger, because the battery had more empty hours to do work in.
What we landed on, for the question of whether to add a battery to a small array, is something close to this. If your tariff has a meaningful peak-to-trough delta and you are disciplined about scheduling, a battery probably earns its keep within ten to twelve years. If you are not, and you are looking at the battery as a "fit and forget" addition to your solar, the case is much weaker and you would be better off increasing the array first.
This piece will be updated when we have a second year of data, or when a meaningful change in tariff structure renders the calculation different. Reader data welcome: hello@energystack.uk.