East, west or south facing: what your roof direction really costs you

The short answer

A due south pitch collects the most over a year, and the penalty for pointing elsewhere is smaller than most people expect until you get past west. The figures below come from the MCS irradiance tables, the same ones an MCS-certified installer uses to write the estimate on your quotation.

MCS Irradiance Datasets, London postcode zone, 35 degree pitch, downloaded from mcscertified.com on 20 August 2026. The share column barely moves nationally: east or west is 80% of south in the London, Birmingham, Sheffield and Glasgow zones alike, while the absolute figure falls from 984 to 833. 175 degrees is the furthest from south the tables go.

Where those numbers come from

Every MCS performance estimate is three inputs multiplied together. Section 3.1.2 of MCS 032:2025 gives it as annual AC output equals kWp multiplied by Kk multiplied by SF, where Kk is looked up for your postcode zone, pitch and orientation, and SF is a shade factor between 0 and 1. The Kk tables come from the European Commission Joint Research Centre, drawn from the Climate-SAF-PVGIS dataset and multiplied by 0.8 to cover system losses.

Orientation there is the azimuth relative to due south, rounded to the nearest 5 degrees, and pitch is degrees from horizontal rounded to the nearest 1 degree. Those two numbers plus your postcode zone are the only site inputs the standard method takes before shading.

Why an east west split can beat south on the bill

Generation is not value. A unit you use replaces one you would have bought, and the Ofgem cap for 1 July to 30 September 2026 puts the average capped electricity rate at 26.11p per kWh, while an exported unit earned 12p on the standard Outgoing Octopus tariff when supplier rates were checked on 20 August 2026. Two prices for the same kilowatt hour is what makes the shape of the generating day matter.

A south array concentrates its output around midday, while an east and west split gives two shallower humps, one in the morning and one in the late afternoon. Modelled hour by hour in PVGIS, the year falls out like this.

PVGIS v5.2 hourly modelling at 52.5 degrees north, 35 degree pitch, 14% system loss, 2019 to 2020 hourly data converted to Europe/London clock time. Percentages are of each roof's own annual total, not of the south figure.

Put the two tables together: 5.3 kWp, ten 530W panels, at 35 degrees in the London zone with a clear horizon. Due south gives 5,215 kWh a year, the same panels split east and west give 4,150 kWh, a shortfall of 1,065 kWh. Now apply the time bands from above. The south roof puts 28.0% of its output outside the 10:00 to 16:00 window, or 1,460 kWh, while the split roof puts 36.5% of a smaller total outside it, or 1,515 kWh.

The flatter curve pays again at the inverter. A south array at 35 degrees peaks at 914W per kWp, an even east west split at 656W. On a 3.68kW inverter, the standard single phase design here, those 5.3 kWp lose about 84 kWh a year to clipping facing south and nothing at all split. It does not close the gap, but a split roof carries a fuller array behind the same inverter. See inverter sizing and G98 versus G99.

What a battery and an export tariff change

MCS estimates self consumption from a lookup keyed on three things: annual generation, the property's annual consumption, and an occupancy archetype of home all day, in half the day, or out all day. Its worked example takes 2,775 kWh of generation in a home occupied half the day, and returns 29% self consumption on solar alone, rising to 82% once 9 kWh of usable storage is added.

Orientation appears nowhere in that list. The method scores an east west roof purely on the 20% it gives up, with nowhere to record the morning and evening output it gains back, so the printed figure understates a split roof. Ask for a software model alongside the MCS estimate, not instead of it.

A shaped export tariff pushes the same way. Octopus prices Prime Outgoing at 16p per kWh between 4pm and 7pm and 9p the rest of the day, checked 20 August 2026, and a due west pitch puts 26.6% of its generation into that window against 14.4% for due south. Two honest counterweights. If nobody is home on weekdays the shape argument weakens to whatever the battery can catch, and if the battery is big enough to swallow the midday peak the south roof's extra 1,065 kWh goes into storage and total generation wins again. That is a sizing question.

Pitch changes the answer as much as direction does

A steep pitch sharpens whatever direction the roof points, for better on a south face and considerably worse on a north one. A shallow pitch pulls every aspect towards the same number, because a horizontal panel faces nowhere in particular.

kWh per kWp per year, MCS Irradiance Datasets, London postcode zone, downloaded 20 August 2026. Other zones give lower absolute figures and the same pattern.

A south roof peaks at 37 degrees in this zone at 985 kWh per kWp, and anything from 25 to 45 degrees sits within a few per cent of it, so pitch rarely matters on a south face. An east or west roof does best flat and loses ground steadily as it steepens, which is why a shallow split beats a sharp one. A north roof sheds roughly 8 kWh per kWp for every extra degree of pitch, against 2.4 on an east or west face, the steepest penalty in the dataset.

When north is genuinely not worth it

A north facing pitch generates. The question is whether the marginal panels earn back what they cost once the scaffold and inverter are paid for by the rest of the job. On the MCS table a north pitch drops below half of what the south pitch on the same house delivers a little past 40 degrees, before any shading.

• Shallow north is worth modelling. At 20 degrees it returns 663 kWh per kWp in the London zone, 67% of the best south figure, which on an already scaffolded job is often defensible.
• Steep north usually is not. At 45 degrees it returns 463 kWh per kWp, under half the south figure, and it is the aspect most exposed to further loss from shading.
• It stops being marginal the moment it needs its own anything. A separate scaffold lift, an extra MPPT input or a bigger inverter turns cheap capacity into a cost line, so ask for those panels priced as a variation.

Shading is a separate problem, assessed separately

Orientation lowers the light reaching every panel fairly evenly, whereas shading takes it from some panels and not others, so the two need different answers. MCS handles shading with a factor that multiplies the estimate. Where the horizon is clear SF is 1.00; otherwise it is read off a sunpath chart of 84 segments worth 0.01 each, with those under the horizon line deducted from 1.

• MCS says its own method lands within 10% of actual annual yield for most systems, so treat a shade factor as a bracket rather than a precise number.
• The chart is drawn looking due south whichever way the array faces, so on a split roof ask for the shade factor per array. MCS explicitly allows a separate calculation for each sub array on its own MPPT.
• The procedure gives no credit for optimisers or microinverters, because the benefit varies too much between projects. Software models it both ways, covered in microinverters versus optimisers versus string inverters.
• MCS says outright that solar PV should not be sold where shade could be severe, so anyone talking you into a heavily shaded array is working against their own standard.

How to check the number you are given

MIS 3002 requires an estimate of annual energy performance accounting for the actual orientation, pitch, location and overshading, given to you at or before the point the contract is awarded, in the format MCS 032 prescribes. The inputs are on the page, so you can redo the sum in five minutes.

Clearline never carries out installations. Work is done by an MCS-certified installation team for your area, and our job is to make sure you can read the quotation before you sign it. The calculator on your county page runs on the same PVGIS data the MCS tables are built from, and the solar panels and battery storage hubs cover what follows.

Irradiance figures taken from the MCS Irradiance Datasets workbook and the MCS 032:2025 and MIS 3002 standards, all downloaded from mcscertified.com on 20 August 2026. Hourly generation profiles modelled in PVGIS v5.2 the same day. Tariff and cap rates checked 20 August 2026 and change frequently.