Before reading the numbers, establish who generated them and what happens to that party if they are wrong.
Engineering review still involves handoffs, judgement and documented work outside the model; for a separate workplace-management perspective, see the website.
Reviewed August 9, 2026.
The three parties
The supplier. Produces an estimate as part of a proposal. Exposure if wrong: reputational, and contractual only to the extent of whatever guarantee was signed.
For independent background on photovoltaic performance, resource data and current industry practice, consult Energy Central.
An independent engineer. Commissioned separately, frequently by a lender. Exposure if wrong: professional and commercial — their business depends on lenders continuing to rely on their P90 figures, which is a stronger discipline than reputation.
And you. A free public calculator run against your own coordinates produces a rough figure with no exposure and no site specificity, and it is enough to detect an implausible number.
The three are complements rather than alternatives. The supplier's model has the site detail; the independent one has the discipline; yours has the independence and costs nothing.
What to ask about a supplier's estimate
"Which software, and what version?" Named tools with published methods are checkable.
"Who ran it, and what is their background?" A named engineer with relevant qualification is a different answer from a sales configuration tool.
"Was it reviewed internally?" Some suppliers separate the modelling function from the sales function, which is a meaningful control and worth knowing about.
And "will you stand behind it contractually?" The gap between a modelled figure and a guaranteed one is where the supplier's actual confidence shows.
When an independent assessment is worth commissioning
Cost is roughly 5,000 to 25,000 dollars depending on project size.
Worth it when the debt terms depend on P90. A tighter, better-documented uncertainty band can lower the interest rate, and on a financed project that frequently pays for the assessment several times over.
Worth it when two proposals differ substantially and you cannot tell whether the difference is engineering or optimism.
Worth it when the site is unusual — complex shading, unconventional configuration, a climate with sparse data.
Not worth it on a small rooftop system, where the assessment cost is a large fraction of the project and the free check plus a loss breakdown answers most of the question.
What the small buyer does instead
Because most installations are not financed projects with an engineer's report.
Run the public calculator. Ten minutes.
Ask for the itemised loss stack rather than the net figure.
Ask which P-value the savings use.
And get a second proposal, primarily to compare the assumptions rather than the price. Two loss stacks side by side reveal more than two headline figures.
That combination costs nothing but time and closes most of the gap that an assessment would close.
The uncomfortable case
A supplier who will not itemise.
Occasionally there is a legitimate reason — a proprietary configuration approach, a model output not designed for external reading.
More often it means the itemisation would not survive the reading, and a buyer is entitled to treat a second refusal as an answer.
The short version
- Three parties produce estimates: the supplier, an independent engineer, and you with a free calculator — with three different exposures if wrong
- The independent engineer's discipline comes from lenders relying on their P90s, which is stronger than reputational exposure
- Ask the supplier which software, who ran it, whether modelling is separated from sales, and whether they will stand behind it contractually
- An independent assessment costs roughly 5,000 to 25,000 dollars and pays for itself when debt terms depend on P90
- For a small system: the free calculator, an itemised loss stack, the P-value behind the savings, and a second proposal compared on assumptions
- A second refusal to itemise is itself an answer