I agree. With my group, sometimes this works, and often it doesn't. What happens in practice is...
- The players 'decide' on one course of action and end up pursuing another which I haven't prepared for (since I prepared for what they said at the end of the last session).
- The players can't decide because they're split or overwhelmed or whatever, and agree to decide via email...and then no emails are sent and no decision made.
- I ask the question too late in the evening after a long session and most players are too tired or need to put little ones to bed, so we resolve to figure it out next session.
When it works, the question works like a charm, but that's maybe a third of the time for my group.
Yes, that fits my experience with asking "What do you want to do next time?" IME most players don't want to decide that until they're actually playing the session. So it's best just to prep for anything you think they're likely to do. Eg if you present 3 clear options at the start of the session they will 99% choose one of those. In fact it's around 80% that after investigating option 1 they will seek to investigate option 2, and 60% that they'll even investigate the third, least-preferred option, given enough time and opportunity. But usually it's best to have more stuff happening before then.
Eg in my Loudwater campaign, the PCs have occasionally discussed investigating the Mountain King, but it's clearly low down on their priorities, so never actually happens, other more pressing stuff comes up.