A 10 kWh battery in a home using 3,000 kWh a year would have saved £239 in London and £266 in North Scotland over the 12 months, a median of 56p a day in London, with a range from 30p on the worst day (27 May 2026) to £2.16 on the best (8 January 2026). Doubling the battery to 20 kWh added nothing, because the limit was how much electricity the home uses when prices are high, not the battery. The full regional table is on the home battery without solar page.
Analysis · Published by SEO Dons Ltd · Updated 14 October 2026
The question, and the data
Agile Octopus charges a different price for every half hour, set the day before from wholesale prices. For a battery owner that makes it the obvious tariff to try, and the obvious tariff to exaggerate: a headline gap between 4pm and 4am looks like a lot of money. The honest way to find out is to run a battery against the prices that were actually charged.
We pulled every half-hourly Agile unit rate from Octopus’s public API for all 14 regions, from 14 October 2025 to 13 October 2026. That is 17,520 prices per region. Prices are as charged, so they include 5% VAT up to 30 September 2026 and 0% from 1 October. We also read Flexible Octopus’s rates for the same period as the flat-tariff comparison.
The prices themselves vary more than the regional averages suggest. In London the cheapest half-hour was minus 11.3p and the dearest 86.7p, against a 12-month mean of 20.3p. The mean of each day’s cheapest three hours was 12.5p, and the mean of its dearest three hours was 35.7p, a spread of 23p a kWh. The evening peak between 4pm and 7pm averaged 35.2p. In North Scotland, prices ran higher throughout: a 24.4p mean and a 40.1p evening peak.
The rules of the test
- The home uses 3,000 kWh a year, shaped by the same seasonal weights as this site’s calculator, with 40% of use between 7am and 4pm, 27% between 4pm and 7pm and 33% overnight and in the evening. It has no solar.
- The battery stores 10 kWh, returns 90% of what it stores, charges and discharges at up to 5 kW, and does at most one full cycle a day. It only discharges to meet the home’s own use; it exports nothing.
- The dispatch is the best the battery could do knowing that day’s prices: for every possible point at which charging stops and discharging starts, charge in the cheapest half-hours and discharge in the dearest, pairing them while 90% of the selling price beats the buying price. Agile publishes prices the afternoon before, so this foresight is available to a real scheduler.
- The comparison. We report the battery’s own saving (the same home on Agile without a battery, against with one), the saving from moving to Agile at all (against Flexible Octopus), and the total.
- Left out: standing charges (the same with or without a battery), export, battery fade and any change in future prices.
What it earned
The battery’s own saving was £239 in London and £266 in North Scotland, and between £245 and £261 in the other regions. Daily, the picture in London is a steady trickle with the odd spike:
| London, 10 kWh | Saving per day |
|---|---|
| Mean | 65p |
| Median | 56p |
| 10th percentile | 38p |
| 90th percentile | £1.07 |
| Best day (8 January 2026) | £2.16 |
| Worst day (27 May 2026) | 30p |
| Days above £1 | 12.6% |
No day lost money, and none earned less than 30p, because the battery stays idle if the gap does not beat the charging loss. In North Scotland the mean was 73p and 18% of days beat £1. The best day in the three regions we examined day by day (London, North Scotland, and Merseyside and North Wales) was the same one, 8 January 2026.
By season, the gain is flatter than for a solar battery. In London the average was about 71p a day from November to March and about 53p from April to August. September 2026 was the best full month at about 86p a day, and May the weakest at about 40p.
The one thing that limited it
The mean of the dearest three hours (35.7p) minus the mean of the cheapest three (12.5p) is 23p a kWh. A battery that stored 10 kWh and cashed in on that spread every day would earn about £2 a day, or £700 or so a year. The back-test earned £239. The difference is not the battery. It is the home.
A home using 3,000 kWh a year uses about 8 kWh a day, and only about a quarter of that falls between 4pm and 7pm. The battery can only be discharged against the home’s own use, so most of what it could earn is capped by how many kWh the home draws in the dear half-hours. On average the battery moved about 5.6 kWh a day, not 10. That is why a 5 kWh battery earned 94% of what a 10 kWh one did for this home, and why sizes above 10 kWh added nothing.
| Household use | 5 kWh | 10 kWh | 13.5 kWh | 20 kWh |
|---|---|---|---|---|
| 2,000 kWh a year | £159 | £161 | £161 | £161 |
| 3,000 kWh a year | £225 | £239 | £239 | £239 |
| 4,500 kWh a year | £310 | £347 | £354 | £354 |
| 6,000 kWh a year | £355 | £442 | £459 | £467 |
Bigger homes, with more evening and overnight use, earn more and use more of the battery. An electric car or a heat pump would change the shape again, but those homes also qualify for other tariffs that this test does not cover.
What it means for a buyer
- Size to your evening use, not to the biggest unit on offer. At Octopus’s April 2026 prices the 5 kWh battery (£3,947) would have repaid itself in about 17.5 years on its own saving, against 22 years for the 10 kWh one (£5,336).
- Count the tariff separately from the battery. Moving to Agile would have saved a London home £105 without any battery, and a North Scotland home £3. A buyer comparing totals should know which part they could have without spending £5,000.
- Ask whether a flat tariff does as well. A battery-owner tariff with a fixed overnight discount, such as British Gas’s Charge Power, can be compared with this on a whole-year bill. The home battery without solar page sets out which tariffs a battery-only home can use.
- Use the calculator for your own use. The savings calculator takes your annual use and region; the back-test says its Agile figures were in the right range, slightly cautious in the year just gone.
The back-test is a tool, not a promise. Prices in the next 12 months will differ, and a battery that does less than perfect foresight would do is worth correspondingly less. But it replaces an argument about averages with a year of real prices, and the answer is: worth something, less than the headline, and limited mostly by how much electricity you use when it is dear.
Questions people ask
How much would a 10kWh battery have saved on Agile Octopus?
£239 a year in London and £266 in North Scotland for a home using 3,000 kWh, across the 365 days to 13 October 2026, on the battery's own contribution. Across the 14 regions the range was £239 to £266. That figure excludes the saving from moving a home to Agile in the first place, which was between £3 and £105.
Does a bigger battery earn more on Agile?
Not for a typical home. For 3,000 kWh a year, 5 kWh earned £225, and 10, 13.5 and 20 kWh all earned £239. At 4,500 kWh a year, 13.5 kWh earned £7 more than 10 kWh, and at 6,000 kWh a year £17 more, because only large homes use enough electricity at the dear times to empty a big battery.
How many times a year does a battery cycle on Agile?
In the back-test, about 204 full-cycle equivalents a year, meaning an average of about 5.6 kWh moved a day out of a possible 10 kWh. The home's use at the dear times limits it, not the battery.
Do negative Agile prices make a big difference?
Not to the battery. London had 484 half-hours priced below zero in the year, about 2.8% of all half-hours, with a low of minus 11.3p, but charging at those prices added only about £6 to the year's saving because the battery was already filling cheaply.
Is the back-test realistic?
It is a best case on one point and cautious on others. The battery knows each day's prices in advance, which Agile allows because prices are announced the afternoon before, but a real scheduler may do slightly worse. It leaves out standing charges, battery fade and any change in future prices, and it uses an assumed shape of daily use, not a particular home.