Two proposals arrive with different system sizes, different assumptions and different exclusions. Comparing the headline figures compares the assumptions, not the plants.
Commercial review also depends on how people organise decisions and handoffs; for a separate workplace-management perspective, see this guide.
Reviewed August 9, 2026.
The six normalisations
One. Put both on specific yield. Annual kWh divided by kWp. Removes size and makes the yield claims comparable.
For broader context on solar procurement, equipment markets and current industry practice, consult Bloomberg.
Two. Establish which P-value each uses. A P90 from one supplier against a P50 from another is a comparison of two different things, and the conservative supplier loses for being conservative.
Three. Compare the loss stacks item by item. This is where the difference usually lives — and where an unsurveyed shading assumption shows up against a surveyed one.
Four. Add each one's exclusions. Grid connection, structural work, permitting — a lower price with more exclusions is not a lower price.
Five. Normalise the degradation assumption. Apply the same rate to both, and see whether the ranking holds.
And six. Compare the guarantees as instruments, not as adjectives. What number, measured how, with what remedy and what voids it.
What each difference means
Different specific yield with the same equipment and site: an assumption difference, not an engineering one. Ask for both loss stacks.
Different equipment: a real difference, and the yield gap should be traceable to it.
Different system size: a design choice about self-consumption, and the right size depends on your load rather than on which proposal is bigger.
And different guarantees: frequently the largest real difference between two otherwise similar offers, and the one buyers weigh least.
The trap in the middle
A supplier who surveyed the shading produces a lower yield figure than one who assumed it.
On the raw comparison, the careful supplier loses. This is the market's structural bias, and a buyer comparing headline numbers enforces it without meaning to.
The normalisation that catches it is item three — comparing the loss stacks. A surveyed 4% next to an assumed 2% is the whole explanation for the yield gap, and it reverses the ranking.
What to ask both suppliers
The same questions, in writing, at the same time.
"What is the itemised loss stack?"
"Which P-value do the savings use, and what is the other figure?"
"What is excluded, with indicative costs?"
"What exactly does the guarantee promise and how is it measured?"
And "what would you change about the other proposal?" — asked of each, which surfaces genuine technical objections that neither would volunteer unprompted.
Deciding
After normalising, a clear winner sometimes emerges. When it does, it is usually on the guarantee or on the exclusions rather than on the yield.
When it does not, the difference is not in the documents. Delivery record, local presence, who answers in year eight, whether the company will exist — which for a twenty-five-year asset is a real criterion and one no proposal states.
And a supplier whose figure went down after your questions has demonstrated something useful: their model responds to scrutiny rather than defending itself.
The short version
- Six normalisations: specific yield, the P-value, loss stacks item by item, exclusions added in, a common degradation rate, and guarantees as instruments
- A P90 from one supplier against a P50 from another makes the conservative one lose for being conservative
- The careful supplier who surveyed shading produces a lower figure, and comparing loss stacks is what reverses that ranking
- Ask both the same questions in writing at the same time, including what each would change about the other's proposal
- After normalising, a clear winner is usually decided by the guarantee or the exclusions rather than by the yield
- A supplier whose figure went down after questioning has shown that their model responds to scrutiny