A customer rings, three years in. The system is not making what you said it would. They forward the PDF you sent them, the one with the annual figure on page four. You go looking for how that number was produced. It came out of a spreadsheet, on a laptop that has since been replaced, by an estimator who has since left.
The MCS redevelopment quietly puts a name to that problem.
The estimate has its own document now
The redeveloped installer scheme is built from four document families. One of them exists, for the first time, as a thing you can point at:
"Pre-sale Information and System Performance Estimate Standards — The technical standards that outline how to carry out system performance estimates and the pre-sale information that must be provided to a customer for an installation. These requirements exist today, but we've restructured and streamlined into their own technology-specific standards to improve clarity."
That is MCS, verbatim, in the CEO's November 2024 letter on the redevelopment. The other three families are Installer Operating Requirements, Customer Commitment, and MCS Installation Standards.
MCS is explicit that this is not a change to the maths. The technical requirements, in their words, "aren't changing… but they have been restructured and streamlined to be more focussed as 'technical truths'".
Nothing about how you calculate the number has changed. What has changed is that the number now lives in a document with its own name.
That sounds like filing. It is not.
From step to artefact
Before the redevelopment these requirements existed — MCS says so plainly — but not as a document of their own. They were part of something larger. The solar PV installation standard is MIS 3002, and MCS has not published which requirements land where under the new structure. What we know is the direction: out of something bigger, into their own technology-specific standards.
A requirement that lives inside a larger standard behaves like a step. You do it, you move on, and the next stage of the job produces the evidence that you did it.
Lift that requirement out. Give it a name. Make it technology-specific. Now it is not a step in a process — it is a standard, and a standard is something an assessor can ask you to evidence on its own terms. Not "show me the installation was compliant, and I will infer the estimate happened somewhere in there". Show me the estimate.
MCS has not published what an assessor will ask for. Neither this post nor anyone else selling you software has read the new standards. But structure predicts scrutiny. Things that get their own document get their own questions.
The surveillance arithmetic
The second half of this matters more than the first. NICEIC describe the new MCS Quality Risk Model as having three surveillance tiers. The strictest is:
"Enhanced assessment frequency – annual assessment… and a minimum of five sites per technology"
MCS has not published how the new standards map onto that assessment schedule, and it would be wrong to tell you they have. But two facts sit next to each other. There is now a standalone, named, technology-specific standard for the performance estimate. And there is an assessment regime that, at its strictest tier, can put five of your sites on the table in a single year.
Five sites means five estimates. Five sets of inputs, five sets of assumptions, five answers to the question "how did you get this number, and is it the number the customer saw".
The rollout is not optional and not distant. Throughout 2026 and into 2027 the redeveloped installer scheme is rolling out, and it affects every MCS certified installer in the UK.
The volume behind this is real. MCS reports 267,032 solar PV installations in 2025. Renbee, who sell DNO application software, report over 120,000 MCS-certified solar PV systems completed in the first half of 2025, up 36% from 2024. Every one of those had a number put in front of a customer by one of the UK's 5,636 MCS-certified contractors. The estimates are not anonymous. They trace back to you.
Independent verification is not a new idea
The trade got here before the scheme did. Writing on ElectriciansForums in July 2021, one experienced installer described EPVS as a scheme that:
"offers independent verification of performance estimates to customers, and will help them in the event that the system does not perform as claimed"
And added:
"This is something the MCS should have been offering since they started."
That is one installer's view, five years old at the time of writing. But it names the instinct precisely. An estimate that only you can reproduce is not an estimate the customer can rely on, and it is not an estimate that protects you when the system underperforms. The redevelopment does not make EPVS redundant or mandatory — it simply moves the scheme's own structure closer to the shape that installer was describing.
What a defensible estimate record looks like
None of the following is a reading of the new standards. It is what any record has to survive: a customer, three years, and a question.
| What the file needs | Why it is hard to reconstruct later |
|---|---|
| The inputs exactly as entered | Roof dimensions and orientation get re-measured, corrected, and overwritten. The version that produced the number is gone. |
| The assumptions behind them | "We assumed the tree comes down" lives in someone's memory, not the file. |
| The shading basis | Whatever method you used to characterise shading, the evidence for it needs to be attached to the estimate, not sitting in a phone's camera roll. |
| Who produced it, and when | Estimators leave. Dates on PDFs get regenerated. |
| What the customer was actually shown | The only document that matters in a dispute is the one in the customer's inbox. |
The last row is the one that catches people.
The gap between the proposal and the file
Most firms have two numbers, and do not know it.
There is the number in your system — the current estimate, re-run after you swapped the panel model, corrected the pitch, or updated the inverter. And there is the number in the PDF the customer received eighteen months ago, which they have kept, and which you have not.
Every re-run widens the gap. The proposal and the estimate record start as the same thing and quietly become two things, and no one notices until someone rings up about yield. Then you are not defending a calculation. You are explaining why your records show one figure and their PDF shows another. That is a much worse conversation, and it is a paperwork failure rather than an engineering one.
This is the part worth designing around now, before an assessor asks. The estimate and the proposal the customer saw should not be two artefacts that agree. They should be one artefact, versioned, retrievable years later, with the inputs still attached to the number the customer read. That is the thing Solar Engine is built to hold.
Before your next assessment
Four things, none of which require you to have read the new standards.
- Pick a job you quoted eighteen months ago. From what is in the file, try to reproduce the annual figure. Time yourself. Whatever you cannot reconstruct is your gap, stated precisely.
- Spot-check five recent jobs — five being what the strictest surveillance tier looks at per technology — and confirm the estimate in your records still matches the PDF the customer holds.
- Decide, now, where the estimate record lives. If the answer is "the proposal PDF", the answer is that you do not have one.
- Watch mcscertified.com for the standards themselves, and do not buy anything on the strength of a blog post about them. Including this one.
The number in front of the customer was always your liability. Now it has somewhere to be filed.