Instead of "when should we add foo," invert it.
Often, the answer is, "effectively zero."
If that's the case, then foo is an inevitability.
That means you can start thinking of how to plan around it now.
"Given that we'll have foo in the future, what small tweaks to our data model or roadmap will make it less challenging when we get to it?"
Sometimes that means pulling forward work a bit, or laying a very small stub that can blossom into the full thing.
The inversion makes what was an amorphous question snap into clarity.