I haven't ever used Skill Challenges in 4e, but I did port the concept across the my recent SWSE campaign with some success. They worked pretty well. My thoughts:
When to Use a Skill Challenge
- Use them only where appropriate. If only one PC can meaningfully contribute, this is probably best handled with a simple skill check. If there's no great drama to the scene, this is probably best handled with a simple skill check, or even just hand-waved. But if they are appropriate, Skill Challenges can be a great tool.
- Use them only where the outcome is in doubt. If the PCs are interrogating an NPC, and you want them to get the information, there's no point in going through the hassle of a Skill Challenge - just have the NPC break down, and move on.
- The consequences of success and failure have to be meaningful. There's a very poor (IMO) Skill Challenge at the start of "King of the Trollhaunt Warrens", where if the PCs succeed then they get where they are going, but if they fail then they are "punished" with a level-appropriate encounter that gives them treasure and XP. And then they get where they're going. (If they were unable to rest after this encounter, or if they were denied treasure and XP from it, that would be better.)
That said, you probably don't want a "succeed or die" Skill Challenge, or even a "succeed or fail the adventure" Skill Challenge. It seems players don't mind 'losing' as a result of a combat encounter (too much, at least), but they have 'losing' as a result of a failed Skill Challenge.
How to Use a Skill Challenge
- Everyone must contribute. I told my players this one up-front - if they had their character just "do nothing", then they'd be assumed to be "getting in the way", and that would count as an automatic failure. Even if their only action was Aid Another, they had to contribute.
(You may think that while the Warlord is haggling with the baron, the best thing your Barbarian can do is say nothing. And it might well be. But if he stands there glowering, and looking like he's about to fly into a homicidal rage at any moment, even that can affect the Warlord's ability to work his charm.)
- Be flexible in the use of skills. See that section in the DMG, the one that talks about 'primary' and 'secondary' skills for the Skill Challenge? Take a big black marker and cross that out - you'll do better to wing it. Have the player describe what they do (which skill they use), then decide if that is directly applicable (primary), or tangentially applicable (secondary).
For secondary skills, I then decide whether the roll will count directly towards the total number of successes, or whether it will simply make other rolls easier. If the former case, I will use a higher difficulty for the roll. In the latter, I will use the normal difficulty, but success will only give a +2 bonus on the next "primary" skill roll.
- Sometimes, they didn't even have to roll. If the players came up with something I hadn't anticipated, and it was a really good idea (as happened twice in my campaign), this was good for an automatic success. Likewise, there were a couple of 'extended' Skill Challenges in the campaign where if the party achieved certain story goals, this was also good for an automatic success.
- I use a flat difficulty number. The campaign I was running ran from 2nd to 7th level, which was basically a "low-level" campaign. For all Skill Challenges throughout, I was using a flat difficulty of 15 for all rolls. This meant that at the start of the campaign, the party failed a lot of rolls, but by the end they were succeeding almost all the time. This gave a distinct sense of progress as the campaign went on - characters became noticably more competent.
Had the campaign gone on much longer, I would have moved to a flat difficulty of 20, but the Skill Challenges would also have become manifestly harder.
- Adjust the success/failure ratio. As written in the DMG, Skill Challenges are too hard. I adjusted this by requiring 3 successes for every 2 failures (rather than 2 for 1, as written). So, it would be 3 for 2, or 6 for 4, or 9 for 6, and so on.
- Narrate 'partial results' for every roll. In a chase scene, if the PCs win a success then you should note that they have found a short-cut, or knocked some boxes in the path of their pursuer, or whatever. If the PCs fail, one of them trips, or a carriage pulls into their path, or they turn down a blind alley, or...
- As with combat, keep it fast. I have a table rule in combat that when your turn arrives, you have to start declaring your first action within 30 second, or you lose your turn. For most Skill Challenges, I use the same rule. Basically, if things are happening 'fast', I expect the players to be decisive. (Of course, I don't use this for every Challenge. It's not appropriate when negotiating passage, when researching in a library, or similar. However, for chase scenes, fast talking, or negotiating an asteroid field, it helps a lot.)
Anything else?
Only this: if you find that Skill Challenges aren't helping your game, don't use them. The game ran just fine without them for 30-odd years, and will run fine without them in the new edition. I like the concept, and find the (modified) mechanics helpful, but YMMV.