When to Use Electricity
The cleanest and cheapest hours where you are
Great Britain’s electricity is not equally clean at every hour of the day, and it is not equally clean in every part of the country at the same hour. The regional model can therefore assign very different average gCO₂/kWh figures to the north of Scotland and the south east in the same half hour.
This page puts the two halves of that question together: the average carbon intensity forecast for electricity in your region in each half hour ahead, and the VAT-inclusive unit rate on the current Agile Octopus product held by this site. It does not calculate an appliance bill or avoided emissions. Pick your region below — the postcode you might use to find it is never stored, and nothing here needs one.
London
Every carbon figure here is a forecast. Regional carbon intensity is modelled from national generation and estimated flows between regions — it is never measured at a region’s boundary, so there is no later outturn that confirms or corrects it.
Cleanest half hour
Tue 12:00
126 gCO₂/kWh
Lowest Agile unit rate
Mon 14:00
26.94p/kWh on Agile Octopus October 2024 v1
The cleanest half hour and the cheapest one are not the same. The cleanest falls beyond the Agile rates published so far, and the cheapest runs at 175 gCO₂/kWh.
Lowest two-hour averages
This compares four consecutive half hours and gives each one equal weight. It is useful for a roughly two-hour continuous load, but it is not the exact cost or carbon impact of a washing-machine programme: cycle lengths and electricity use within a cycle vary.
Lowest forecast average intensity
Tue 11:30 – 13:30
128 gCO₂/kWh on average
Lowest average Agile unit rate
Mon 13:30 – 15:30
170 gCO₂/kWh on average, 27.21p/kWh
The first 48 forecast half hours
- FromgCO₂p/kWh
- 05:30202g32.10p
- 06:00207g36.84p
- 06:30216g37.80p
- 07:00205g39.59p
- 07:30273g40.17p
- 08:00268g39.12p
- 08:30261g39.23p
- 09:00255g35.68p
- 09:30252g33.66p
- 10:00242g33.05p
- 10:30238g31.89p
- 11:00231g30.60p
- 11:30222g28.84p
- 12:00211g28.41p
- 12:30207g27.92p
- 13:00201g27.55p
- 13:30190g27.41p
- 14:00175g26.94p
- 14:30162g26.94p
- 15:00151g27.54p
- 15:30149g27.91p
- 16:00159g41.59p
- 16:30176g46.12p
- 17:00185g49.98p
- 17:30193g52.23p
- 18:00210g55.61p
- 18:30214g56.00p
- 19:00212g41.03p
- 19:30212g40.83p
- 20:00209g40.38p
- 20:30207g37.44p
- 21:00203g37.21p
- 21:30200g31.52p
- 22:00197g33.24p
- 22:30197g30.30p
- 23:00195g—
- 23:30192g—
- 00:00185g—
- 00:30185g—
- 01:00183g—
- 01:30186g—
- 02:00185g—
- 02:30191g—
- 03:00194g—
- 03:30200g—
- 04:00198g—
- 04:30189g—
- 05:00190g—
The carbon forecast runs 95 half hours ahead, and the table above shows its first 48 published slots. 60 later slots fall beyond the Agile rates published so far and show a dash. The held price series runs to Mon 23:00; Octopus says the following day’s rates are normally published between 4pm and 10pm.
Cleanest half hour ahead, by region
| Region | Cleanest | gCO₂/kWh | Dirtiest | Lowest Agile rate |
|---|---|---|---|---|
| Eastern England | Wed 04:00 | 129 | 387 | 28.47p at 14:00 |
| East Midlands | Mon 14:30 | 186 | 308 | 26.94p at 14:00 |
| London | Tue 12:00 | 126 | 273 | 26.94p at 14:00 |
| Merseyside and North Wales | Tue 03:30 | 20 | 99 | 29.99p at 14:00 |
| West Midlands | Tue 13:00 | 98 | 238 | 28.47p at 14:00 |
| North East England | Tue 06:30 | 9 | 18 | 28.47p at 14:00 |
| North West England | Mon 10:30 | 18 | 40 | 28.47p at 14:00 |
| Southern England | Tue 12:30 | 121 | 272 | 28.47p at 14:00 |
| South East England | Mon 13:30 | 112 | 195 | 29.99p at 14:00 |
| South Wales | Tue 13:00 | 289 | 392 | 29.99p at 14:00 |
| South West England | Tue 14:00 | 67 | 389 | 31.51p at 14:00 |
| Yorkshire | Mon 15:00 | 90 | 173 | 26.94p at 14:00 |
| Southern Scotland | Mon 05:30 | 0 | 0 | 28.47p at 14:00 |
| Northern Scotland | Mon 05:30 | 0 | 0 | 33.03p at 14:00 |
Southern Scotland, Northern Scotland are published at 0 gCO₂/kWh in every half hour of this window. That is the publisher’s own forecast rather than missing data — a region with nothing published shows a dash in these columns instead.
These are VAT-inclusive unit rates for the current Agile Octopus product held by this site. They do not include the daily standing charge. On a flat tariff the unit rate does not change by half hour, so the price comparison does not apply. Contains Carbon Intensity API data from National Energy System Operator, licensed under CC BY 4.0. Tariff data from the Octopus Energy public API. Energy Review is not affiliated with, or endorsed by, Octopus Energy.
The carbon figures describe the forecast average mix in each region. They do not measure the marginal emissions caused or avoided by one appliance, so use them to compare lower-intensity times rather than to calculate an emissions saving.
Past cheap windows
Is the cheap window always at the same time?
This is where the cheapest three hours have actually fallen on past days, not a prediction for tomorrow. The forecast for the day ahead is above.
Nearly always overnight, but not at a fixed hour. The cheapest three hours began between 10pm and 5am on 89.2% of days, and in the middle of the day on 8.6% — usually a sunny one, when solar pushes the midday price down.
- 12am–3am22.3%
- 1am–4am7.4%
- 2am–5am6.7%
- 3am–6am8.2%
- 4am–7am4.5%
- 5am–8am0.4%
- 6am–9am0.1%
- 7am–10am0.1%
- 8am–11am0.1%
- 9am–12pm0.4%
- 10am–1pm1.9%
- 11am–2pm3.7%
- 12pm–3pm2.5%
- 1pm–4pm0.1%
- 2pm–5pm—
- 3pm–6pm—
- 4pm–7pm—
- 5pm–8pm—
- 6pm–9pm—
- 7pm–10pm—
- 8pm–11pm0.6%
- 9pm–12am1.3%
- 10pm–1am—
- 11pm–2am39.7%
Only tariffs whose price actually changes through the day are counted. On a flat tariff every half hour costs the same, so there is no cheapest window to find - and flat tariffs are the majority of what this site stores, so counting them would invent a pattern. Based on 155,232 tariff-days over the last 365 days, with 221,875 flat-tariff days left out.
Why the carbon figure is always a forecast
National carbon intensity gets checked. A forecast is published ahead of time, the generation is metered, and an outturn figure replaces it — which is why the live electricity page can show what actually happened rather than what was expected to.
The regional figures work differently, and the difference matters. No meter records the carbon crossing the boundary between one distribution region and the next. What is published instead is a model: national generation, apportioned by what is connected where, adjusted by estimated flows between regions. There is no later measurement to validate it against, so it is not later replaced by a measured regional outturn. Treat these numbers as a guide to when the modelled average mix is lower, not as a measurement of the emissions caused or avoided by a household load.
If you are on a flat rate, only half of this applies
The price column is Agile Octopus, a tariff whose unit rate changes every thirty minutes with the wholesale market. Most households are not on one. On a standard variable rate — the tariff the price cap applies to — a kilowatt hour costs the same at three in the morning as at six in the evening, so moving a wash saves nothing at all on the bill.
Moving use to a lower forecast-intensity half hour aligns it with a cleaner modelled regional mix, but this average-intensity feed cannot quantify the marginal emissions consequence. It is also worth knowing the gap between the lowest and highest Agile unit rates before switching: whether a household saves depends on its actual half-hourly consumption, standing charge and the flat tariff used for comparison.
When households actually use it, measured
Everything above this point is about what the grid is doing. This is about what households are doing. SP Energy Networks and SSEN each release an aggregated half-hourly profile from their low-voltage networks, but their coverage and time conventions differ, so the two panels below stay separate rather than becoming one average.
Each is a limited window, and neither is a forecast or a model. They are what the meters recorded, within the scope stated on that panel.
SP Energy Networks feeder profile
Ayrshire and Clyde South and The Wirral only. Two parts of two distribution licence areas — Ayrshire and Clyde South inside SP Distribution, and the Wirral inside SP Manweb. Neither is a licence area in full. This is not Scotland, not England and not Great Britain, and no other network operator publishes an equivalent open measurement. Anywhere else is not covered, which is not the same as having no data. September 2025, published in January and February 2026 and not updated since. This is a measured month in the past, not a live feed.
Ayrshire and Clyde South
Part of SP Distribution. About 18,004 metered households across 1,001 street circuits, averaged over 23 days.
2.51× as much electricity in the busiest half hour of a weekday as in the quietest
Watt-hours per metered household per half hour. Weekday use peaks at 17:30 and is lowest at 05:00, as published — see the note on the clock below.
The Wirral
Part of SP Manweb. About 65,030 metered households across 1,774 street circuits, averaged over 28 days.
2.45× as much electricity in the busiest half hour of a weekday as in the quietest
Watt-hours per metered household per half hour. Weekday use peaks at 17:30 and is lowest at 05:00, as published — see the note on the clock below.
How many homes have a smart meter at all
The whole of SP Distribution and the whole of SP Manweb — a much wider area than the consumption sample above. Context for it, never a denominator for it.
- SP Distribution
- 60.6%1,187,350 active smart meters across 1,958,114 properties; 59% of them second-generation
- SP Manweb
- 63.7%942,682 active smart meters across 1,479,145 properties; 63% of them second-generation
About the clock. SP Energy Networks declares no timezone for these datasets and September is British Summer Time, so the clock times here are either UTC or local time published under a UTC offset — an hour apart, and the daily shape fits both. The ratio between the evening peak and the overnight minimum is the same figure under either reading, which is why it leads.
About the denominator. The denominator is smart meters, not homes. A feeder carries households without a smart meter whose use is not in these figures at all, so this is what the metered subset used and not what the street used. Feeders with fewer than this many reporting meters are excluded before anything is aggregated. SP Energy Networks suppresses below five; this is deliberately stricter, because a five-household group on a row that also carries a substation coordinate is not an anonymising crowd. It shifts the mean by about 3% in Ayrshire and Clyde South and under 1% in the Wirral, and the excluded feeders are disproportionately rural.
Contains data from SP Energy Networks, licensed under CC BY 4.0
SSEN measured feeder profile
Southern Electric Power Distribution and Scottish Hydro Electric Power Distribution only. SSEN's two distribution licence areas combined — southern England and north Scotland. Two of the fourteen GB licence areas. This is not Great Britain, not Scotland and not England, and the two areas cannot be shown apart: the published usage file carries no geography beyond the operator name, and the only thing that would separate them is a feeder-to-postcode lookup this programme deliberately never reads. A rolling window of recent days, refreshed daily. The publisher releases each day about three days later.
SSEN’s two licence areas combined
Southern Electric Power Distribution and Scottish Hydro Electric Power Distribution. About 2,067,742 metered households across 74,537 street circuits, averaged over 48 days.
1.87× as much electricity in the busiest half hour of a weekday as in the quietest
Watt-hours per metered household per half hour. Weekday use peaks at 18:30 and is lowest at 06:00, on this feed’s measured UK wall-clock basis.
About the clock. SSEN declares no timezone and publishes each stamp with a Z suffix, but the values are UK wall clock rather than UTC — settled on the two clock-change days, where a UTC series and a local one give different numbers of distinct labels and the data gives the local answer both times. The times here are therefore the household's own clock and need no conversion. The clock-change days themselves are excluded, because one label there means two different half hours.
About the denominator. The denominator is smart meters, not homes. A feeder carries households without a smart meter whose use is not in these figures at all, and it carries non-domestic connections too, so this is what the metered subset used and not what the street used. Feeders with fewer than this many reporting meters are excluded before anything is aggregated. SSEN suppresses below five and drops those rows entirely; this is deliberately stricter, because SSEN also publishes a feeder-to-postcode lookup that makes a five-household group locatable. It moves the mean by about -1.8% and excludes 21% of feeders but only 6.8% of metered households.
Contains data from Scottish and Southern Electricity Networks, licensed under CC BY 4.0
SPEN’s two published sub-areas show the shape particularly clearly: household electricity use roughly doubles and a half between the small hours and the evening peak, and both areas peak in the same half hour despite being three hundred kilometres and two licence areas apart. That is part of the demand shape the grid has to be built for.
That measured demand shape explains why system operators value flexibility, but it does not prove the carbon effect of moving one wash. The regional forecast above can identify lower average-intensity half hours, while the cost of balancing the grid reports a separate system-wide cost.
Why one region is cleaner than another
Wind farms, nuclear stations and gas plants are not distributed evenly across the country, and electricity does not travel freely between regions. When Scottish wind is abundant, the transmission network sometimes cannot carry all of it south — the constraint that shows up as the largest single line in what balancing the grid costs. The result is a region that is very clean while another, at the same moment, is burning gas.
Agile rates share a national wholesale input but also include regional and time-dependent factors. That is why their pattern can resemble the carbon forecast without either series determining the other, and why the lowest values need not occur in the same half hour.
Frequently asked questions
Is the carbon figure a measurement of what my region emitted?
No, and it never becomes one. National carbon intensity is published as a forecast and then confirmed by an outturn once the generation is metered. The regional figures have no outturn at all — they are modelled from national generation plus estimated flows between regions, because nobody meters the carbon crossing a distribution boundary. Everything on this page is a forecast, including the parts about the past.
I am not on a half-hourly tariff. Is this page any use to me?
The carbon half is still a guide to when the forecast regional grid mix has lower average intensity. The price half is Agile Octopus, which changes every thirty minutes; on a flat tariff the unit rate does not change by time, so the displayed price comparison does not apply. The regional intensity figure is not a marginal-emissions model, so it cannot tell you the emissions caused or avoided by moving one appliance.
Why do the cheapest and cleanest half hours often differ?
They are set by different things. Agile unit rates use wholesale costs plus regional and time-dependent pricing factors. Regional carbon intensity is a model of the average generation mix and estimated flows for that region. The two series can move together, but they are not measuring the same thing and need not identify the same half hour.
One region shows 0 gCO₂/kWh. Is that data missing?
No — it is the figure the publisher issued, and it arrives on the same feed at the same time as every other region. Where we hold no forecast at all the figure is a dash rather than a zero, and the table names any region published at zero right through the window so the two cannot be mistaken for each other. It is still a modelled regional figure with no outturn, so nothing later confirms or corrects a zero either.
Why do prices stop partway through the chart?
Agile prices are calculated from a formula related to wholesale costs. Octopus says the following day’s rates are usually published between 4pm and 10pm. Until those rates arrive, later price cells are shown as dashes rather than filled in. A separate warning appears if a rate is missing inside a price window that has otherwise been published.
Does running appliances overnight really help?
There is no dependable “overnight is cleanest” rule, which is why the page shows the current regional forecast. The forecast compares average grid carbon intensity between half hours; it does not calculate the marginal emissions effect of starting a particular appliance, and its lowest-intensity time may be during the day.
Related reading
Written by
Rob Gibbs
Hi, I'm Rob, and I run Energy-Review.co.uk. I initially started this project in 2018 when I was looking to switch energy suppliers and found there wasn't a website that provided simple, data-backed reviews on all the suppliers available. Since then, I have spent a lot of time (too much, some may say!) looking at all publicly available data about each supplier and writing reviews using this information. These reviews are updated as regularly as possible, and any data is backed up by a source where necessary. I have also started writing guides on various energy-related topics, which hopefully you will find useful. If you find any issues, please use our contact form to let us know.