You have four tabs open. Each is a solar design tool. Each demo draws a roof in ninety seconds and each one looks fine. None of them show what happens months later, when somebody asks where the performance estimate came from and you are looking at a PDF you cannot regenerate.
The roof is not the hard part. Drawing roofs is a solved problem and has been for years.
An installer posting from Dorset asked this on ElectriciansForums in January 2023, working through an MCS calculation:
Am I correct in thinking I can just use easy pv to do this calculation if I put in correct figures. Or would they expect me to have a more in depth knowledge on this subject.
That is the industry in two sentences. Design software has quietly become compliance software. Nobody sold it that way and nobody certified it that way, and the person signing the certificate is the one carrying the difference.
Where these tools grew up
There were 5,636 MCS-certified contractors in the UK in Q4 2025, and between them 267,032 solar PV installations during that year — both figures MCS's, as reported by a secondary source. That is a real industry. It is also a small one — small enough that no global software company was ever going to build its data model around British paperwork first.
You can see this in where each product's audience actually is. Similarweb models website traffic from panel and clickstream data. These are modelled estimates of who visits a website. They are not search volumes, not counts of installers, and not market share. A tool with 87% Australian traffic can be an excellent tool and can have UK customers. Read the table as evidence of centre of gravity, nothing more.
| Tool | Traffic origin (Similarweb, modelled) | Market origin |
|---|---|---|
| Easy PV | ~96.7% UK | A UK product serving the UK market. |
| OpenSolar | ~22% UK; also large in the US and Australia | Global. Publishes and documents a UK-specific MCS calculator. |
| Aurora Solar | Predominantly US; the UK is not in its top five countries | Centre of gravity is the US market. |
| Pylon | ~87% Australia | Centre of gravity is the Australian market. |
None of that is a criticism, and it is worth being exact about why. A tool built for California solves California's problem. A tool built for New South Wales solves that one. Neither was ever going to grow a native understanding of MIS 3002 or a G99 application, because neither of their first thousand customers needed one. That is not neglect. That is where the customers were. Building for your own market is the right thing for a software company to do. It just means you should know which market that was before you commit your business to it.
Worth saying: OpenSolar publishes a UK-specific MCS calculator and documents it. A global platform that bothers has spent money on your paperwork on purpose.
Software has already won, which is the problem
The Only Panels installer podcast ran an episode on 1 October 2025 covering, among other topics, "Why software matters more than hardware in solar" and "The rise of OpenSolar, convenience, and solar being 'bought not sold'".
Convenience is the mechanism. The tool that produces a proposal in ten minutes wins your Tuesday. Then it quietly wins the shape of the job. Whatever your software makes easy becomes what your business does; whatever it does not model, you do in a spreadsheet at nine in the evening and nobody ever sees it again. This is not a procurement decision about drawing tools. It is a decision about which parts of your process become invisible.
What to evaluate on instead
Three things, none of which appear in a demo unless you ask.
Does the output survive an assessment. The solar PV installation standard is MIS 3002, and whatever your tool prints has to sit alongside it. The document your assessor reads is not the document your customer signed. Ask to see the compliance pack, not the proposal. Vendors demo the pretty one.
Does the performance estimate become a record. In the redeveloped MCS Installer Scheme, the performance estimate is no longer a line item buried in the technical rules — it sits in its own family of documents, "Pre-sale Information and System Performance Estimate Standards". MCS's own framing is that the technical requirements "aren't changing… but they have been restructured and streamlined". Read that structure honestly: the number you show a customer before they sign is being treated as a regulated artefact. So the question is whether your tool can retrieve, three years from now, the exact estimate a customer was shown, with the inputs it used — after the software has updated its irradiance dataset twice. A figure that regenerates against today's data is not a record. It is a recalculation wearing a record's clothes.
Does anything carry survey data into the DNO application. Renbee, who sell DNO application software, report that "over 60% of applications submitted via Connect Direct are escalated for manual review, while only 0.6% are automatically approved". They also report that installers are "asked for new data points, such as estimated maximum demand post-installation or the cut out rating, without sufficient guidance or training to work this out". Take the interest into account — they sell the fix, and the sample is their own platform. Then notice which data points they name. Maximum demand. Cut-out rating. Both are things you establish on a survey, and the survey is where your design tool already is, sat on your knee in somebody's loft.
The chain nobody spans
Line it up. Design, then performance estimate, then DNO application, then MCS certificate. One job. Four artefacts. Each one restates facts established by the one before it. The array is specified in the design, restated in the estimate, restated in the G99, restated on the certificate — four times, by hand, by you, with four chances to disagree with yourself.
No tool we can find spans that chain. Solar Engine does not span it either. We work in the middle of it, and we would rather write that down than let a demo imply otherwise.
Which gives you a test. If a vendor says their product covers all four, ask to see the certificate coming out of the far end, populated, nothing re-keyed. Then watch their hands.
Five questions for a demo
Ask these of every vendor. Ask them of us as readily as of anyone else.
- "Show me the pack an assessor sees, not the proposal a homeowner sees." If the answer is the same document, the tool has one audience and it is not your assessor.
- "Retrieve the performance estimate this tool generated for a job two years ago, with the inputs it used." Watch whether they retrieve it or recalculate it.
- "Which fields of a G98 or G99 application does this populate, and where does each one come from?" "None, and here is why" is a perfectly good answer. A shrug is not.
- "What happens to my designs when I stop paying." Export format, ownership, and whether last year's jobs come with you.
- "Who at your company has read MIS 3002 this year." Not a gotcha. It separates a product with a UK owner from a product with a UK reseller, and you will hear the difference in the pause before the answer.
What to do first
Take the last job you finished. Put its four artefacts side by side and count how many times you typed the same array size into a different box. That number is your integration cost. It is the only benchmark here that measures your business rather than somebody's demo.
Then book two demos: the UK-native tool, and one of the global ones. Ask the five questions in the same order, and write down where each conversation stalls. The stall is the information.
Nobody in this market has earned your trust with a render. The question was never which tool draws the prettiest roof. It is which one still has your back in three years, when somebody asks where a number came from — and for now, the honest answer from every vendor, ours included, is: partly.