I'm new to DMing and I've been planning my first adventure (for D&D 5e). However when it comes to non-combat situations I don't know when it makes sense for a skill check to be used or when the NPC would just reply.
To answer this question, we need to first consider how D&D handles actions. It usually goes something like this:
GM describes situation.
Player declares action.
GM determines whether player needs to roll check.
Player rolls check if necessary.
GM describes outcome.
The step you are having difficulty with is step 3.
Ability checks tell us whether the character is able to perform a certain task, whether that task is jumping over a ravine, or persuading the king to give him command over a few soldiers. Some GMs like to use ability checks to determine something about their world. That is to say, if you fail to pick the lock, that means that that particular lock is too difficult for your character (i.e. you can't just try again). Others like to see each ability check as just an attempt by the character (i.e. you can try again).
Regardless of your approach, there is one thing in common: a possibility of failure. If there is no possiblity of failure, then there is no need for a check, since whatever the player rolls, their character will still succeed. For example, if the player wants to persuade the innkeeper to let them buy a meal, there is no chance that the innkeeper will say "No, I don't want your money." Therefore, in that situation, no check is needed, you just let them succeed immediately.
What if the chance of failure is 100%. That is to say, the character has no chance of succeeding at all. For example, perhaps a player wants to try to jump across a 100m chasm - that's impossible. In that case, whatever the player rolls (even a natural 20) should fail. There is therefore no need for a check, you just tell them that they failed immediately. However, when this happens, there is usually some misunderstanding. If the character is trying to jump across a 100m pit, it's pretty safe to say that that player thought that the pit was much narrower. For this reason, a good policy is to explain that such an action would be impossible before adjudicating the result: "You do realise that that pit is 100m wide right? There's no way your character can jump across."
With the above in mind, let's expand step 3:
Determine whether the action has a chance of success. If it does, move on to step 2. If it does not, tell the player that the action is impossible (when appropriate). If they still want to try, describe the consequences of their failure.
Determine whether the action has a chance of failure. If it does, move on to step 3. If it does not, tell the player that they succeeded, and describe the consequences.
Determine how difficult the action is. There is a nifty little table on page 238 of the DMG to help with this. This table is also on the official DM screen. if you find yourself choosing "very easy" or "nearly impossible", then you might want to reconsider steps 1 and 2 to see if you should call for a check at all.
Tell the player to roll the check. Whether or not you like to tell them the DC in advance is up to you.
Once you get practised with this, it becomes almost instinctive. For now though, you might want to think about it in steps as I outlined.
I'm now going to address the idea that checks should only be called for when failure is interesting.
Let's imagine a situation in which the character are climbing up a ridge. Suddenly, the bit of rock that they are clinging on to falls off the cliff face. If they fell to the bottom, they would immediately die (a TPK). It could be argued that failure here is not interesting, so you should just let them somehow save themselves. However, it should be obvious that this is almost impossible. At this point, the players begin to feel that your world is not consistent. That can really harm your game, as players base all their actions on the idea of a consistent world.
Another example would be the tight-rope above pillows. The question is this: Why did you include it in your adventure? If you want the characters to succeed on something, then make it so easy that they can't fail to succeed first time.
What you have there is a problem with your adventure design. You should design your adventures (or run modules) in such a way that, if there is a chance of failure, then it is interesting. If you are running a murder mystery, don't have the whole thing hinge on one clue, otherwise what happens if the players never even go to the place where they can find it? Instead, sprinkle clues liberally, all pointing to the same place. The players will miss one and misinterpret another before finally working it out (although that's somewhat of a cliche).
Finally, if for some reason the player's have failed the entire adventure, let them. If they always succeed somehow because you fudge things, there is little sense of accomplishment when they do succeed. But, you must have a way to signal to them that they have failed, otherwise they will keep banging their heads against the wall.
As a final addendum, I strongly recommend that you check out The Angry GM's how to GM section. There is so much good stuff there about how to run games, adjudicate actions, run combat, etc. I have found it immensely useful. He seems to be pretty much the only person on the internet who codifies GMing into easy to follow steps, rather than assuming that all GMs know exactly what they're doing.
Skill checks are for when failure is interesting
If you want to know whether or not you should call for a skill check, ask yourself 'If the PCs (player-characters) didn't succeed, would the consequences be interesting?' If the answer is 'No', then just have them succeed. If the answer is 'Yes', then you might want to call for a skill check.
From your question it looks like you're talking mostly about social situations, so let's take a look at those. If your players are talking to an NPC, and they ask that NPC for a piece of information that they need, try to determine if that information has any particular value to the NPC, if the NPC even knows the information is important, or if the NPC has a reason to be untruthful. If they're asking for the location of the local king's garrison, that's probably common knowledge that any local would have, so there's no real reason to hide information about it. Furthermore, if you made it a check and the PCs failed, they could just ask the next person they see the same question, and continue that ad nauseum until they finally passed the check.
However, if they're asking a Baron for the password to let them into the local religious cult's midnight ceremony, this is privileged information that the Baron would know is valuable. In this case, failure is 'interesting' because with the password they can just walk into the ceremony, but without it they will need to sneak or punch or bribe their way in, all of which carry additional risks. Asking for a check here will make the PCs approach this social interaction strategically, which is really what the skill check system is all about.
The same line of reasoning can be applied to all checks. Perception checks to notice a bandit sneaking into the camp is interesting (does the bandit surprise the party, or is the party prepared). Perception checks to notice a key clue in a mystery are not interesting, because failure just means the party chases its tail until you finally decide to give the clue to them. An acrobatics check to walk across a tightrope suspended between two trees is interesting if the rope is 50 feet in the air or if it's over a pit of snakes, but it's not interesting if it's only 5 feet in the air, or if it's over a herd of fluffy pillows.
Use a skill check when:
- There is opposition
- There is a meaningful chance of success or failure
- Role-playing is not sufficient to the task
Opposition means someone is trying to get an NPC to do something against their basic nature or against their basic interest. Trying to buy something from a merchant is not going against their basic interest. Trying to buy something from a Capulet merchant when you are a Montague might be. Trying get a merchant to sell you something at or below cost certainly is.
This sort of opposition might be active or it might be passive-- an NPC doesn't need to know he is in a skill check situation to be resisting something. It could be something as simple as a guard's or even a bystander's tendency to notice something weird.
Meaningful chance of success keeps you from rolling check after check in inappropriate situations. No, the guard of the jail you are in is not going set you free just for asking nicely. No, the king's previously established highly loyal brother is not going to betray him.
Of course, there are situations which could change that. The guard of the jail might be amenable to a bribe, the king's loyal brother might have something ugly to blackmail him with. But those are situation specific, and shouldn't be conjured up by the die roll; rather, the die roll doesn't even become appropriate unless and until those factors are in play.
And the opposite is true as well: A 20th level barbarian who just punched out a dragon probably does not need a check to intimidate the local barkeep. Unless that barkeep is a living trope who happens to be a retired 20th level wizard.
Role playing not sufficient is a catch-all, and might be controversial. When might role playing not be sufficient to the task? Often when either the GM or the player or both do not want to go through the full details of something. I am famously allergic to role-playing haggling scenes: Just roll the damn dice. I would absolutely and entirely draw the line at any scene approaching "roughing up the prisoners" and cut straight to a skill check. (And give my players the side-eye above and beyond.) On the player side, it is reasonable to get a little hesitant or awkward at, say, a seduction attempt.
Note, of course, that all of these guidelines are subjective, and that your interpretation of these guidelines is both a result of, and will inform, the style of game you run. (As will the difficulty modifiers you choose to impose based on the situation.)
Social skills are tricky. I do recommend studying the PHB section on Using Ability Scores (Charisma) where it talks about social skills - that can tell you which skill is appropriate, but not whether to call for a roll in the first place.
Ultimately it's a matter of taste, so you need to work with your players to develop a policy. Some GMs roll very little, and base social outcomes on your actual dialogue. But not everyone is good at improvising under pressure. To me, being able to play a character who's better than their player at eloquent persuasion is as important to the RPG experience as being able to play a character who's better than their player at, say, swordplay.
So what I usually do is say "Sure, you can just roll for that if you want to, but give me a general idea of your approach and what you're trying to get at." Then if they have a particularly good idea, I give them a bonus to the roll, regardless of out-of-character stuttering or whatever. Of course if people want to do the full dialogue, that should be an option too.
But really, anything can work, as long as you're fairly consistent and up front with your players. You may also be interested in the Angry DM's take on when to call for a roll at all; the consequence of failure in most social situations is that a bad first impression will linger, meaning you can't just try again until you get it right. On the other hand, if your plot requires a friendly NPC to fill the players in on what's going on, have them volunteer rather than making the plyers roll for it. At most, they could roll to see how long it takes them to find someone who'll fill them in- but even then, only if it matters because time is precious.
Since I don't see this type of opinion already...
Simple answer: Always, assuming the situation would require a skill check.
Two things are noteworthy, the first is that you have a list of known skills for your character, as well as ones you are proficient with. That being said, if the task at hand fits into one of those skills then that check should be used.
For more dynamic and social situations, skills should be used anytime you are attempting to get something outside what the NPC would freely give (being that anything they would freely give does not require a check since any such check would pass automatically). One thing to note though, you have your skills and ability scores. Each ability is tied to some sort of feature for your character. If you think something should require a check then decide which might be appropriate to roll for your situation (e.g. a conversation dice roll not covered in the skills would likely be Charisma).
Exceptions: (DM Discretion)
- Anytime you need to roll where the result is either instant pass or instant fail (regardless of roll) I generally recommend not bothering with the roll. Simply say something like "you attempt to intimidate him but he just laughs at you".
- If a check (e.g. persuade the king to send help to a village) is critical for your campaign, have your characters roll it normally, but give them more ways of achieving the same thing. Perhaps one of the court advisers already knows of the problems and believes you without needing the roll, and attempts to convince the king after you leave. If your game hinges off of a single roll then it should probably be planned better (or with more options).
- Role-playing is a style which a lot of people prefer to use rather than rolling. This can be used in lieu of rolling in most situations if you wish to reward your players (or if your game is more role-playing based).
- Give bonuses for creativity (personal favorite).
Use social skill checks when you start thinking "Hmm" (or, "Yeah, right")
D&D is a role-playing game, you want to encourage good role-playing - that's where the fun is. But a character's social skills should also be useful (if you want player ever to choose them) so use skill checks to let social skills provide the character success when the player's role-playing falters.
Imagine a player trying to talk her way through a challenge (gaining admittance, haggling a better price, whatever). If the player has a great idea (maybe something that refers to something else in the game) then go ahead and grant an automatic success - that's a reward for role-playing. This will seem quite natural to do.
If the player comes up with a half-baked idea, then go ahead and say, "Um, make a roll for that" (using the appropriate skill).
Sometimes players will "come up empty" and just say they'll try to talk their way by the guard. That's OK, just call for the roll. The player has effectively passed on their chance for an automatic success (or advantage) from role-playing.
Don't impose too many penalties
Sometimes after a role-playing attempt a DM will call for a roll, but with a bonus or penalty. That is fine, but be judicious in doling out penalties - use it only when a player says something that would be blatantly offensive or obviously false. You don't want to discourage the players from attempting to role-play.
Use social skill checks when the player starts it. This will likely be because the player has built a socially adept character, and wants to roll some big rolls. That's fun, so let them.
Example 1) Your player is playing Sam the Sneak, a rogue that is REALLY good at lying, and he wants to use dice. Let him roll Sam's Bluff check, roll your Sense Motive check, and play out the results of your checks accordingly.
This also applies when the player on the receiving end. If the player says they want to roll Sense Motive, let them. Roll your Bluff check, then play out your results.
Example 2) Your player is playing Wilfred the Wise. He doesn't think Sam the Sneak is on the up and up, and Wilfred has a super duper Sense Motive, so he says he wants to roll and confirm that Sam's a liar. Fine, roll your Bluff and play out the results.
In both cases, whether you roll dice in the open or not depends on how you and your players like your game to go. Maybe at your table, it doesn't cause a problem if everyone knows that the NPCs are lying, because everyone has just as much fun playing his character's confusion as he would when his character sees through the lies. In that case, roll heads up. Maybe the players prefer to be guessing themselves. In that case, you should hide your dice. Maybe even roll for the player, as well, for maximum uncertainty.
It's important though that the players bring the dice out, not you. If you're playing Sam the Sneak, and the player decides that his character doesn't believe what he's saying, he doesn't have to roll to back that up. The character might be wrong, and Sam is telling the truth this time, but a player is the master of what his character believes and thinks, even if he's wrong. Unless there's magic involved.