So when I set out to make my cube, I have to decide: how large? how
powerful? how balanced? And the underlying question beneath all of it
is: how fun? If the card combinations or theme result in fun games,
people will want to play it, if not, you'll be home alone. The size of
the cube is fairly key, as you need to have a minimum amount of cards to
support a certain amount of players drafting. Most successful first
time cube builders start small, usually in the 360 card (minimum for an 8
player draft) or 450 card (minimum for a 10 player draft) range. Typically, the larger a cube gets, the more iffy, borderline, or bad cards get included. If you want your cube to have iffy cards, then you're free to build as you like. I decided that I wanted my cube to have as few undesirable cards as possible, without being game-breakingly overpowered. I wanted the first pick in the first pack of a draft of my cube to present some tough choices. When you crack a pack in a regular pack and see one great card, one good card, and a bunch of fodder, your choices are easy. If someone was to crack a pack of my cube, I wanted them to see 10 great cards, 3 good archetype supporting cards, and 2 situational cards that might be great parts of a combo.
I spent a lot of time studying a lot of successful cube lists. I spent a lot of time reading Tom LaPille's archives, Adam Prosak's writing, and especially Usman Jami. I studied what kinds of balance they designed into their cubes, balance between colors, card types, converted mana cost, and took note of the cards that seemed to keep appearing consistently across cubes that I knew to be of good quality.
I would really recommend anyone undertaking a cube project to spend a good amount of time planning their cube, and doing research. Researching Magic theory and design can be very difficult and frustrating. Always consider your sources. Anyone can login to MTGSalvation and throw up a list that's completely untested and consists of whatever popped into their head at the time. If you have attended Friday Night Magic anytime in the last 5 years, you may have realized there's a really wide spectrum of player skill and card evaluation ability out there. Getting a feel for how well an author understands the game and cards behind it will save you a lot of time in the long run.
It was also in doing this research for the size and prospective cards for my cube that I discovered a unique cube concept that I decided to make my own: the split card cube. Split cards
are cards with two card faces. A split card is literally has two separate cards on a single card. Several of them appeared in the Dissention and Time Spiral sets. A classic example of a split card is Fire/Ice:

So Usman Jami created a cube where all of the cards were split cards. One half would be something like Mirran Crusader and the other could be... Wrath of God. I thought this could make for some really interesting card interactions. Your hand size is essentially doubled, and drafting a balance of cards becomes even more key. Other strategies take on interesting dimensions, like graveyard recursion. If you cast a spell on one half of a card and it's in your graveyard, cards that interact both with spells and permanents in your graveyard will have twice as many targets.
Deciding on a split card cube does present a unique set of challenges as well. The effective card pool for the cube basically doubles in size. I decided on a 450 card cube, sufficient for 10 people to draft it. Going with a split card cube means that I needed to select 900 cards! Still, I thought the idea of the split cube presented a unique enough experience that would make people want to draft it more. We'll see if I've bitten off more than I can chew.