What I think leads to a successful session is first you need some type of objective. You need obsticals to reach that objective that are not too hard and not too easy. You then need some type of reward for achieving that objective. The reward has to be good, something that the players want- wealth, power, a cool new toy, whatever- but its got to be something they want.
If the game is supposed to be social then every character also would need some type of social thing to do. Maybe one character wants to raise his own army or castle or such. Socially, he might have to deal with underlings and such. A common reward for him would to be gaining more followers, and a common penalty for failure would be loosing followers for mistakes he made while roleplaying with the underlings. One player might want to become some type of wisened, all-knowing sage. So then he's got to do lots of social roleplay with people to gain knowledge--its not exactly something you can beat out of someone (except for Highlander). He might gain some awesome book or a new mentor as a reward, failure he might have insulted a mentor so much that they become an enemy.
On objectives, I think you need to have multiple ways of getting past them. Easier ones should have at least 3, harder ones should have more. They should also have about 1/3 to 1/2 of additional ways of "beating" them that don't involve success, but failure. Best case senerio, the failure choices should feel slightly less as good as the right choices. You could mix this up though, throw in a clue- has to be fairly blatently a clue, perhaps not when they first hear it but for sure when it comes time to make a choice. The clue would be something that tells the players that making the "less likely" choice as being the right choice, instead of the normal setup where the right choices are the best ones and the failure choices are the worst. Having multiple right choices is also important, especially for the hard puzzles, because players might not always think along the same lines of logic as you did inventing it. This increases the chance of them figuring it out in a reasonable amount of time. If a puzzle/riddle is taking too long to figure out, namely when its very apparent they are making no progress toward achieving it, players can get frusterated and bored.
In the case of failure, I think there are a few good concequences that don't involve death. Take d&d for example, a character losing a valuable piece of equipment can be a huge blow. Also I think its important to make sure a player feels like they deserved what they got. A player could get really upset at losing their favorite sword or finding out their family's multimillion dollor corperation just got bought out from under them. If its obvious that this was a direct result of something they themselves did by making a bad decision or whatever, there's a lot less chance of them getting really upset (they will always get a little upset, you just don't want too upset--if they didn't get upset at all, then there's a good chance they probably don't even like playing their character. If that ever becomes really obvious, I would probably ask them if it was true and suggest making a new character if they wanted). You could also have just a general loss of wealth in the case of a player with lots of money, or with loss of reputation. It could even be something simple like now they have to take the "long way around". This might actually be best to write down a list of general ideas and save it somewhere to refer back to later. Maybe even have it with your other DM notes as a type of chart you could refer to in-game if you need a quick idea.
Character development would be very important. This is of course player-dependent. Its good to ask players how they'd like their characters to develop, perhaps one long-term goal and several short-term goals; the long-term goal could change during the campaign depending on what happens, so I would put over half of the emphisis on goals toward the short-term ones. Short-term goals could also be droped because of character development, or achieved, either way probably within 2-3 sessions with new short-term goals replacing them. I think the goals are important because character development is dependent on the PC's player, not the DM. Making them think of goals is about the best way you can wing them along and it gives you some insight on the way they feel about their character.
