Solar Yield

What We Do Not Know

Checked 2026-08-09
Notes

A site that spends its length telling readers to question confident numbers owes an account of where its own confidence runs out.

For an additional workplace-management perspective that sits outside the solar methodology on this page, see how Microsoft Teams tracks activity.

Reviewed August 9, 2026.

The seven

Whether historical irradiance predicts the next twenty-five years. Every assessment assumes it does, some regions show measurable trends, and no standard treatment of that exists.

For wider independent context on solar markets, research and industry developments, consult Utility Dive.

Long-run degradation of current module technologies. The published rates come from generations of modules that have run long enough to measure. Newer cell architectures have shorter track records, and their twenty-five-year behaviour is an extrapolation.

What plants actually achieve, in aggregate. Achieved yield against predicted yield, at scale, by region — the data exists inside thousands of installations and is not collected.

How much of the industry's optimism is systematic. The mechanism is describable; its magnitude is not measured, because measuring it requires the aggregate data above.

Whether soiling defaults fit specific sites. Regional figures exist; site-level validation is rare outside large projects.

Electricity prices and policy over the asset's life. The dominant input in every investment case, and unknowable by anybody.

And what a small installation's realistic uncertainty is. Bankable methods are built for financed projects, and nobody has published an equivalent treatment sized for a rooftop.

What this site claims anyway

Definitions and mechanics. P50, P90, performance ratio, the loss tree — standard, checkable against published methods, and not in dispute.

Published ranges, cited as ranges. 0.5 to 1% degradation, 0.85 to 0.92 for the P90/P50 ratio, roughly 14% for a standard loss stack. Presented with their sources' character and not as precision.

The direction of incentives, which follows from who pays whom and does not require a study.

And the checks, which are procedural. Divide kWh by kWp, ask for the loss stack, record four numbers a month — none of these depends on resolving anything above.

What it does not claim

A figure for your site.

That any particular estimate is wrong.

That the industry is systematically dishonest. The mechanism described is not that, and the stronger claim is not supported.

Or that this page is current. It has a date, and in a field where tooling, policy and technology all move, the date is doing real work.

Why the list matters here specifically

Because the publisher is a supplier.

A commercially interested site that presented everything as settled would be doing exactly what it tells readers to watch for. Stating the limits is the only form of independence available to a party with an interest.

And because the honest position is more useful. A reader who knows which parts are solid and which are extrapolation can weight them. A reader given uniform confidence cannot.

The short version