Proving an underperforming solar system
The commonest complaint and the hardest to evidence. Weather moves a year by a few per cent, not by a third — so a small shortfall proves nothing and a large one points somewhere specific. Here is how to tell which you have, using data you already own and figures we publish free.
Start with the size of the gap, because it tells you what kind of problem you have.
Year-to-year weather variation across our 25 UK reference locations is 2.4% to 4.1% at one standard deviation. So:
A shortfall under about 10% is probably weather, soiling, or an optimistic estimate at the margin. It is not usually a case.
A shortfall of 20% or more is not weather. It points at shading that was not assessed, a fault, or an estimate that was never achievable for that roof — and all three are actionable.
Step one: what should it produce?
You need a modelled figure for your roof, not a national average.
Take your system size in kWp — from the quote, the MCS certificate, or the sum of the module ratings.
Find your nearest location, pitch and orientation in our published dataset. It gives annual output in kWh per kWp for 25 UK locations at five pitches and five orientations.
Multiply. That is your unshaded modelled annual output.
Apply a shade factor if the roof is shaded. MIS 3002 expresses this as annual output = kWp × Kk × SF, where SF is 1.00 for a clear horizon. Our shading page gives modelled factors for fifteen scenarios.
For orientation: at 35° facing south the UK range is 692 to 1,042 kWh per kWp. If the figure implied by your original quote was above about 1,050, it was above anything our dataset achieves anywhere at that pitch and orientation — which is a specific and checkable point.
Step two: what did it actually produce?
Use metered generation, not the app’s estimate. A generation meter or the inverter’s lifetime total is the number. Monitoring apps sometimes report modelled or estimated figures.
Use a full year. Solar is severely seasonal — our irradiance data shows two thirds of annual output arriving between April and September, and the June-to-December ratio ranging from 3.6× on the south coast to 14.9× in Shetland. Judging a system on a winter quarter tells you nothing.
Record it monthly if you can. A month-by-month comparison localises the problem: a system that tracks the model in summer and collapses in winter is behaving differently from one that is uniformly low.
Step three: read the gap
| Shortfall against model | What it usually means |
|---|---|
| 0–5% | Normal. Inside year-to-year variation |
| 5–10% | Probably weather plus soiling, or a slightly optimistic estimate |
| 10–20% | Worth investigating. Shading, partial fault, settings, or a poor estimate |
| 20–35% | Something specific is wrong, or the estimate was never achievable |
| Over 35% | A fault, a substantially unassessed obstruction, or a system not doing what it was sold as |
The temptation in a dispute is to claim the maximum. Resist it.
A complaint asserting a 40% shortfall that turns out to be 12% loses credibility on everything else. A complaint that says “the modelled figure is 3,676 kWh, I have metered 2,740 kWh over a full year, that is 25% short, and here is the model I used” is very hard to wave away.
Precision is the persuasive thing here, not size.
The common causes, and who is responsible
Shading that was not assessed. The most frequent real cause. MIS 3002 requires a shade factor in the performance estimate, and where the MCS method was used and SF is below 1.00, you are entitled to the sunpath diagram used to calculate it. Ask for it. If no shading allowance was made on a roof that is obviously shaded, that is an estimate problem, not a weather problem.
One panel dragging a string. Shading or a fault on a single module can reduce the whole string. Optimisers or microinverters mitigate it; a plain string inverter does not. If the array crosses a shaded region and no module-level electronics were fitted, that is a design question.
Inverter fault or settings. Including an export limit configured tighter than agreed — see export limiting. Check what limit is actually set.
Soiling. Real but usually modest in the UK, and it accumulates gradually rather than appearing suddenly.
Wrong orientation or pitch assumed. Quotes are sometimes modelled from an aerial photograph. MIS 3002 requires a specific note where the site was assessed remotely, warning that actual performance “may be significantly lower or higher” — if that note is on your paperwork, the estimate was never a survey.
An estimate that was never achievable. Check the implied kWh per kWp against our dataset. This is a mis-selling question rather than a fault.
Getting an independent second opinion
For anything you intend to pursue, this is the step that matters.
Use someone with no interest in the outcome. Another installer quoting for remedial work is not independent. An MCS installer or an independent surveyor charging a fee for an inspection report is.
Ask for specifics: measured string voltages and currents, inverter fault logs, a shading assessment with a sunpath diagram, confirmation of the export limit setting, and an opinion on whether the installation matches the design.
Ask them to compare against a modelled figure and to say which cause they consider most likely.
Get it in writing. A verbal opinion is not evidence.
Assembling the claim
The original estimate and everything you were shown. The performance figure, and the assumptions attached to it — or the absence of them.
A full year of metered generation, monthly if possible.
The modelled comparison, naming the source. Ours is free, published and citable.
The independent report.
The paperwork that should exist — MCS certificate, shade factor and sunpath diagram, building regulations certificate, DNO notification. Missing documents are concrete failures that are easy to evidence.
Then: installer first, then the consumer code’s dispute service, with a section 75 claim in parallel if credit was involved. The sequence is on consumer codes and dispute routes.
If the system is fine and the estimate was not
This is a common and awkward outcome. The panels work, the inverter is healthy, nothing is faulty — and the system still produces far less than you were told it would.
That is not a defect claim. It is a misrepresentation claim, and the evidence is the original sales material rather than the hardware. Section 50 of the Consumer Rights Act may also make the figure you were given a term of the contract. Both are covered on mis-selling.
The distinction matters because pursuing the wrong one wastes months. A working system that was oversold is a paperwork case, not an engineering one.
Sources
- Photovoltaic Geographical Information System (PVGIS) 5.3
- MIS 3002: The Solar PV Standard, issue 6.0
- Consumer Rights Act 2015, section 49
Contains public sector information licensed under the Open Government Licence v3.0.
Sorry to hear that. What was the problem?