What a solar installer knows that a solar investor does not

Model an array and you get a curve: nameplate output, a degradation rate somewhere near half a percent a year, an inverter replacement around year twelve, and a tidy internal rate of return. Every number in that model is defensible and the model is still usually wrong, because none of the things that actually go wrong appear in it.

The failure list, in order of how often it happens

Almost nothing that fails is a module. Modules are the most tested, most commoditised, most warranted object in the system, and they mostly do what the datasheet says for longer than the datasheet promises.

What fails is everything around them. Roof penetrations that were sealed on a cold day. Cable routed across a sharp edge that took four years to chafe through. String designs that put an inverter at the edge of its window so that one hot summer clips a percent off every afternoon. A shading survey done from a satellite photograph in March, before a neighbouring tree had leaves on it.

Nobody has ever lost money on a solar array because the physics was wrong.

Why the survey is the whole job

The single highest-leverage hour in a solar project is somebody standing on the roof. Not because the roof is complicated, but because that hour is the only point at which the structure, the penetrations, the cable route and the year-round shading are all being decided by one person who can see all four.

Every hour after that is executing a decision made in that hour. Every euro saved by skipping it comes back with interest, usually in year six, usually as a monitoring alert that nobody is contractually obliged to answer.

The part that is an actual business

Here is the structural oddity. The industry is arranged so that the party who makes the decisions with the longest consequences — the installer — has the shortest relationship with the asset. They are paid on completion. The party holding the twenty-five-year exposure is usually meeting the array for the first time in a spreadsheet.

Anyone willing to stand on both sides of that gap is in a slow, unfashionable, extremely defensible business. It does not scale the way a platform scales. It does get better every year, which almost nothing else in this sector does.