The feature you shelved because the cost per request killed it goes back on the roadmap
Things get cut because the arithmetic does not work, not because nobody wanted them. Change the arithmetic and the decision that killed them changes with it.
Every roadmap has a graveyard, and most of it was not killed on merit. Somebody worked out the cost per request, multiplied by the volume, and the feature stopped being a product decision and became a budget one.
That arithmetic is not a law of nature. It is a function of what each request costs to serve, and that is the number we move.
We are careful about what this figure is. It describes our engine answering our benchmark population, and it is published so that a stranger can check it against a running system rather than take our word for it. It is not a promise about your traffic — that is what a shadow run is for.
The useful version of this news is not the percentage. It is that a backlog item you wrote off is worth costing again.