I want to try to give a bit of analysis on the existing answers, which were the two things my mind also jumped to when I saw the question.
The flaming property adds 1d6 fire damage to weapon damage. Torches, as improvised weapons, add 1 fire damage to their weapon damage. So we have a range of +1 to +1d6 damage; 1d6 averages 3.5, so it is about 3½× as good as the +1.
Personally, I feel that an improvised fire arrow isn’t likely to be much better than an improvised torch, and probably not nearly as good as true magic. But it probably is a little better than the torch; after all, the fire is, ideally, getting plunged into the target. So for consistency, I’d probably rule the damage to be +1d3 (average 2) or +1d4 (average 2.5).
That said, consistency isn’t the only concern. There’s also a desire to reward clever players, and possibly to improve mundane access to alternate damage types. After all, non-magical characters have painfully few options for dealing with, e.g., swarms. So from that perspective, I probably would go with +1d6 fire damage. But, I might also improve the flaming property itself, now both for consistency (should be better than an improvised fire arrow) and for improving player options (it’s a bit lack-luster on its own, perhaps).
Gwet Ovar Hier!
For the rope+arrow thing, I’d probably rule it an impromptu harpoon, from Frostburn, though with arrow damage and proficiency requirement, and an improvised weapon penalty on the attack roll (−4). Now, the harpoon does not actually allow you to drag enemies to you (and I feel that the arrow would slip out or break, or the rope would break), but it does stick in opponents, and you can hold the rope to prevent them from moving further away from you. The enemy can remove the arrow, but it wastes an action and deals extra damage (though I’d probably reduce both of these with respect to the harpoon, unless it was specifically a barbed arrow or something).
Mostly, this tactic probably just wouldn’t work, and while I would give the players something for rule of cool, I don’t think I could justify very much here. I would suggest that they might be able to work on making a real version of what they tried to improvise, and that might work better.