[Anaraxes]

"Succeed with an unrelated benefit" is a kind of critical success. You get htat by rolling really well. The difference is that instead of the benefit being "in line" with the success, it's orthogonal. When the hero attempts to motivate the villagers and rolls a crit success, you could express that result as "convincing all of them", as in the OP, but you could just as easily express it as "convince most of them and learn about the spells". If such events aren't frequent enough, then you could lower the bar on the MoS requirement from "win by 10" (for a crit) to some lesser number.

"Succeed at a cost" occurs on what the dice say is a failure. To turn that into a success, you need to "buy" a modifier sufficient to move the result back to a success. That is, it's a negative MoS that you need to overcome. That Margin of Failure tells you how big the cost has to be to succeed despite it. You might award that margin to the bad guys, who get to use it for free on one of their rolls. Or you can turn it into a narrative effect, which would require working out a scale of examples for what a +2 change would be, or a +5, or a +10.

"Taking a consequence" to succeed means putting some other currency on that chart: you might choose to suffer some FP loss, HP loss, or temporary or even permanent Disadvantages to buy that success. You might rate the CP involved in the feature, and have a discount factor based on how easy it is to repair (rest for the FP, heal for days on the HP, full value for taking a permanent Disad, etc)

Another way to look at the mechanic would be to treat it as a sort of negative Deceptive Attack (on any sort of roll). Rather than take a penalty to your roll to get some benefit (reduced defense), the player chooses to claim a bonus on the roll (to ensure success) by taking a penalty elsewhere. This might not be as directly related to the action as in a DA. It also might be a narrative effect, as in the "some villagers get themselves killed" case.

The problem with narrative effects is of course stopping the game to debate what happens and whether it's serious enough to buy the success; that is, how many MoS points any invented situation might be worth. Again, you can work out some typical examples, but since the game situations have a lot of variety, you'll have some unavoidable GM-player negotiation.

The scope of the effects probably ought also to be related to the scope of the roll. Sometimes die rolls are for small things -- one hit in a combat. Sometimes they're big -- persuade the king to declare war. The consequences likely to be similar in scope. While it might be okay to burn some FP as a desperate defense to get a couple of points on one defense roll, it would might a lot less sense to merely burn a couple of FP to win a heated debate that changes the course of the setting. A bigger stake might be called for.