An independent yield assessment is a document produced by an engineer who is not selling you the plant. It costs real money and on the right project it is the cheapest item in the budget.
Teams reviewing yield assumptions can keep the engineering model separate from the record of review work; the website is one option for documenting time and handoffs.
Reviewed August 9, 2026. Costs are indicative and vary by market and project size.
What it costs and what it buys
Roughly 5,000 to 25,000 dollars depending on project size.
For independent background on photovoltaic performance, resource data and current industry practice, consult PVcase.
It buys a defensible P90 — one a lender will accept, with a documented uncertainty analysis behind it.
And it buys a second opinion on the assumptions, which is frequently the more valuable half. An assessor with no stake in the sale prices shading, soiling and availability differently from one who does.
When it pays for itself
When debt terms depend on P90. A better-documented, narrower uncertainty band can lower the interest rate or raise the debt capacity, and on a financed project that arithmetic usually settles the question immediately.
When two proposals differ materially and you cannot tell whether the difference is engineering or optimism.
When the site is unusual — complex shading, an unconventional configuration, a climate with sparse irradiance data, a bifacial or tracker design whose gain assumptions you cannot verify.
And when the sum at risk is large relative to the assessment cost, which is the general form of all three.
When it does not
On a small rooftop installation, where the assessment is a large fraction of the project cost.
When the free checks already answer the question. Specific yield against a public calculator, an itemised loss stack, and a second proposal compared on assumptions cost nothing and close most of the gap.
And when you will not act on the result. An assessment that arrives after the contract is signed is an expensive document.
What to specify when ordering
The exceedance levels required — P50 and P90 at minimum, P75 and P95 where the lender asks.
The irradiance dataset and its length, and whether ground validation is required.
A full loss tree, itemised.
An explicit uncertainty analysis with its components separated — inter-annual variability, data, modelling, technology — rather than a single combined figure with no derivation.
The degradation assumption and its basis.
And the plane of array, stated, so the figures are not quietly horizontal.
What to do with it
Compare it against the supplier's model, item by item. The differences are the conversation, and they are usually in the judgement items rather than the physics.
Ask the supplier to respond to the differences. A supplier who can justify their figure against an independent one has strengthened their position; one who cannot has told you something.
And use it as the contractual reference if you can. A guarantee measured against an independently derived P90 is a different instrument from one measured against the supplier's own model.
The intermediate option
A desktop review rather than a full assessment.
Cheaper, faster, and it checks the supplier's model rather than building a new one — the assumptions, the dataset, the loss stack and the uncertainty treatment.
It will not produce a bankable P90 and it will find an optimistic loss assumption, which for many buyers is the whole question.
The short version
- Roughly 5,000 to 25,000 dollars buys a defensible P90 and a second opinion on the assumptions, which is often the more valuable half
- It pays for itself when debt terms depend on P90, when two proposals differ inexplicably, when the site is unusual, or when the sum at risk is large
- It does not pay on small rooftop systems, when free checks already answer the question, or when the contract is already signed
- Specify exceedance levels, dataset and length, an itemised loss tree, a component-level uncertainty analysis, the degradation basis, and the plane of array
- Compare item by item against the supplier's model and ask them to respond to the differences
- A desktop review is the cheaper option: it checks the supplier's assumptions without producing a bankable figure