The question is the order.
At portfolio scale the question is not whether a roof works. It is which roofs, in what order, and on what basis that order can be defended. Boards approve sequences and budgets, not one building at a time.
The estate
Thirty installer proposals are thirty quotations, not a ranking. Each carries its own yield, cost, tariff, and self-consumption. Most of those assumptions are never stated.
At portfolio scale the question is not whether a roof works. It is which roofs, in what order, and on what basis that order can be defended. Boards approve sequences and budgets, not one building at a time.
The usual path is to collect proposals, building by building, often over months, often from different suppliers. It feels like diligence. You end up with a stack, and the question still open. A roof that looks better may simply have been quoted by someone more optimistic.
Irradiance sources and loss assumptions change, and often no basis is stated at all.
Scaffolding, grid works, structural strengthening, and cable runs vary by quote.
Small differences here move payback by years.
Self-consumption is often a flat share, not modelled against the load.
Change any one and the order of the estate can change with it.
You cannot back out a comparison from figures whose assumptions were never stated. Rebuilding thirty proposals is modelling the estate anyway, slowly, from worse inputs.
Every building on one model, banded by return, with the assumptions stated once. The question “why is this roof above that one” has an answer. The value is consistency, not a claim of perfect precision on any single roof. A firm installer quote still replaces a modelled cost band when you procure.
At five buildings the assumptions can be checked by eye. At fifty the errors look like differences between roofs. That is where collecting proposals stops being a way to order an estate. Not because anyone acted in bad faith. Proposals are built to price work, not to be ranked against each other.