The idea of free will is that an entity with it isn't fated to one choice vs. any other, right? They might have a preference for one or several options over others, and they might be open to outside persuasion, but that's no guarantee of their future decision. In short, a past decision where free will played a role could have gone another way.
The problem is there's no way to check that. We can speculate on past decisions, sure, and brainstorm about future decisions, but the number of possibilities we can actually observe is either 0 (before the point of decision) or 1 (after).
Even if we speculate a technology (or other method) by which we may view multiple universes, wherein one (initial) decision (semi-arbitrarily chosen to be interesting) was resolved several different ways, that doesn't prove any involvement of free will. Heck, this scenario constructs an infinitely-branching (perhaps even merging) multiverse wherein no timeline is more clearly valid than another (actually, that might be mistaken, but saying so necessitates a metaphysical mathematics I've never heard of); that is, if you don't even need free will to make all the decisions at once, you don't need free will for any one of them.
What's needed is a way to replay a timeline to check whether anyone behaves differently when they all start with perfectly identical memories and perfectly identical circumstances. But even then, there's no way to get completely outside and say that any resulting differences weren't fated.
I dunno. It's something I had to mull over. Maybe I can get some puzzles up this month.