I've started looking at this a bit differently.
Most discovery questions are designed to confirm what you already suspect.
There's a different kind worth trying: "What would have to be true for this not to work?"
It surfaces concerns the prospect hasn't said out loud yet. Usually the ones that actually matter.