This is a cost centric perspective. I wish I could find the YouTube talk where Merck switched to a metric of "cost of not implementing" IE: how much revenue do you lose per day by not implementing. They found out of 100s of projects, that about two were 3 orders of magnitude larger, yet they delayed shipping those projects so they could get more in a release cycle, and diverted resources. From that new perspective, it was obvious that they should stop doing everything except those two projects. From the same talk, Microsoft found a third of projects were revenue negative! (Better to just not do them at all)
The point of the talk was ultimately that the revenue of projects usually dwarf the cost, to the extent that if a project is worth doing, it is clearly so whether it takes 3x or even 10x the time estimated.
The point of the talk was ultimately that the revenue of projects usually dwarf the cost, to the extent that if a project is worth doing, it is clearly so whether it takes 3x or even 10x the time estimated.