I think you're right that the cabinet is pre-built in a warehouse, and arrives on-site in a pretty-much working state.
However, I guess the "lost" money isn't in the value of the cabinet (or the included kit) - as that can be re-deployed elsewhere.
We've also seen that some cabinets can end up left unused for quite a while, so BT obviously aren't overly worried about the lost income from a small number of cabinets - especially if the big picture has sufficient custom coming in.
That leaves us with the fact that the real lost money is in the civil engineering work to prepare the site - build a concrete plinth, and set the ducting within it.
While pulling fibre is probably the variable with the biggest distance involved, it is also within BT's control more readily. And where the fibre is part of a spine, then there is more than 1 cabinet to bring into the equation.
However, I don't think the fibre itself amounts to the biggest variable - even if there are duct blockages - in terms of cost.
I think the biggest variable comes down to power again - but this time *not* in the distance that the power has to be connected, nor in the leadtime that an order has to be placed.
Instead, I think that the cases where an installed cabinet suddenly becomes so unviable it is worth removing (and yet was once viable in the spreadsheets) come down to a huge price spike in supplying power. It might be because power isn't available where the planners thought it was, but I bet the real reason is when transformers or substations need a (pretty costly) upgrade, and the charge is planted 100% at BT. In such circumstances they might as well wait until someone else foots the bill. A property developer perhaps, or the council.