So, there are a few things you need to accept as baselines for a Fantasy heartbreaker. Characters will probably have some manner of race/class combination, there might or might not be some skills or feats (or feat-like things), maybe some levels, and perhaps most importantly, stats.
Statistics are one of the first things you think of when you look at a character's sheets, so much so that "Stats" is conversationally synonymous with the character's sheet. With that in mind, that's the first mechanical thing I'm thinking about. The choice of stats can be a simple thing, but it's worth some time and thought, specifically to determine how many stats to use and what they should represent.
Setting aside special stats, the classic split is between the mental and the physical. It's entirely possible to get by with just those two stats, but fairly dull. Still, they make a good foundation, and how you split them says a lot about the game.
The first split is probably one that it may seem like I'm overlooking: Social. Only slightly less common than the 2 way split is the 3 way split between physical, mental (knowledge) and social (spiritual). This 3 way split is the basis of Tri-stat and Storyteller/Storytelling stats. In two-way splits fold social under mental, usually when social interaction doesn't get a lot of mechanical support.
D&D is actually kind of fascinating to look at through this lens. From a physical/mental perspective, it's a split set, 3 & 3. From a three way perspective you get a 3:2:1 split, which is probably more reflective of the real priorities of the game. Contrasted with White Wolf, which has a 3:3:3 split.
I probably want to have at least a little social support, so let's go with a 3 way split as the foundation.
Given that, I could stop at 3 stats, but that's still pretty dull, so the question is how to split them.
The first option is to split them consistently. The Storytelling system does this by splitting each category into three subcategories that correspond roughly to power, finesse and resistance. This is pretty intuitive for physical stats, but not necessarily so much for mental, and it's definitely jarring for social stuff.
The second is to look at how they'll be used, and split it like that. Unfortunately, I'm starting from scratch here, so I have no real answer to that. Still, I may take a lesson from that.
Another possibility is to look at stat systems I like. Rolemaster left its mark on me, for example, but to be honest I'd be hard pressed to remember all the stats it used, so perhaps not the best example. Dragon Age has a pretty nice set of 8, though one of them (magic) is a special stat, with the rest having a 3:3:1 split, with social getting the short end of the stick with the nicely named "Communication" stats. However, arguably it's actually 3:2:2 depending upon how you use Cunning.
Now, here's where we start getting into the more trivial-seeming areas of the decision making process. First, I don't want names of stats that sound too stupid or too technical. Second, I'd like an even number of stats within a given range, so 4, 6, 8 or 10. The good news is that these limits actually clarify things: 4 is too few, and 10 is more than I really want to try to keep in mind, so 6 or 8.
6 would be ideal. It's easy to remember and the fact that it's D&D's number speaks well for it. Eight would be a function of necessity if I want to flesh out the possibilities. Part of the problem is that Strength, Dexterity and Endurance really are hard to go without, and that's half of a 6 point spread right there.
Assuming we're going to go with 8, I want at least a 3:2:2 split. This prospect leaves me waffling a bit, and this has me inclined to steal a page from Dragon Age, and shuffle magic back into the deck. This will introduce some complications - making the magic stat mean something for non-magical characters will be important, but I have a thought on that - and it rounds things up to 8, nice and tidy.
So, we've got 4 stats already: Strength, Dexterity and Endurance (maybe Agility rather than Dex) and Magic. That leaves 4 slots, and I think I'm going to go with a structured split (active/passive). For mental, that's say Cunning and Willpower, and for social I'll steal from our own playbook and go with Rapport and Composure.
[EDIT: Discussion in comments has convinced me that the mental pair is actually Intellect and Focus.]
This almost certainly seems like overthinking a fairly simple decision, especially this early in, but it's important to think about why you make a choice so you're prepared to change it later on if you find the design suggests one thing, but your earlier decision is still saying something else. Right now, this spread of stats isn't saying a lot (intentionally) but it's suggesting an emphasis on the physical, classical sensibilities, the presence of magic, and a non-trivial role for mental and physical activity. Will those still be the case down the line? if it's not, then by understanding why choices were made, it's easy to understand how to change them.
1 - Such as for things like magic.
2 - There's an instinctive desire to consider a symmetrical distribution to be automatically superior, but it's an unfair assumption. The distribution should reflect the priorities of the game - if it does that well, it's the right distribution.
3 - Theft continues to be fun, so I'm going to steal from the Liavek novels and tie magic to luck. No, I'm not sure what that means yet, but it seems like an idea to start with.
4 - Compare with my non-Amber stats (Force, resolve, grace, wits) or a non-stat system, like Aspects or the Smallville iteration of Cortex.