I've just started DMing (D&D-5e).

I have a question about the Insight mechanic. I've formulated this problem using D&D but the system is otherwise irrelevant (I'm talking about every equivalent mechanics), I'm just asking about an idea. However, I'll gladly read D&D-only answers :)

How does Insight work? Is it about expectations of players or actual situation (bluff or not)? Is it ok to lie in response to a low result as long as it's what the player expects?

Let me present an example. Let's say I have a paladin in my game who is evil (but the players don't know that). He asks the players to kill some evil people, his servants. The paladin's motives are to kill the servants because they were disobedient as well as kill players because they are threat to him. This request is probably suicidal because the players probably can't accomplish the task. The paladin knows that.

One of these servants is cruel, bloodthirsty and totally opposite to the paladin's behavior. After a fight with this servant, the players talk to him. He says that the paladin is his leader.

That seems absurd to the players, so they roll an insight check and get a 5, which is far below the DC.

What to do?

  • Lie to them that this guy is bluffing? He's not but it fits in expectations of players.

  • Don't roll at all in the first place because no deception has occurred.


10 Answers 10


A successful Insight check should reveal useful information to players, and an unsuccessful check should emphasize uncertainty

In the PHB, the Insight skill is defined as

Your Wisdom (Insight) check decides whether you can determine the true intentions of a creature, such as when searching out a lie or predicting someone’s next move. Doing so involves gleaning clues from body language, speech habits, and changes in mannerisms. (PHB, Chapter 7, Using Each Ability, Wisdom, Wisdom Checks)

My guiding principles with Insight are:

  • That it reveals information about the immediate situation being examined, and not necessarily about the world itself
  • Success suggests a keen understanding of the circumstance, while failure indicates poor understanding

I feel it is important to include both elements.

In your specific example, I would present these elements as the bloodthirsty servant being sincere in his belief that he works for the paladin on a successful Insight check. Players should not learn the true state of the world (definitively discovering the relationship between these NPCs just based on evaluating the servant's claim) from examining what the servant says.

In the case of a failed Insight check I would narrate the outcome as a lack of information, rather than being certain that incorrect information is definitely accurate. The failure of insight just means that they have no particular understanding of meta-information about the servant and the assertion the servant has made. The narration I favor would emphasize that-- "he doesn't seem to be obviously lying to you, but you can't get a good read on him at all."

I recommend not forbidding checks because there is no dishonesty to discover. This directly reveals to players the same information they would get on a successful Insight check, but without having to roll. When something seems off to players, or they want to double-check information they receive, they should be encouraged to try to find out via their characters' ability to examine what they know and perceive.

I suggest not giving false information on a poor roll for similar reasons: players will know they rolled poorly (unless you use a hidden-roll mechanic), and so telling them definitive information pretty clearly marks that information as unreliable. I also advocate not directly lying to players more broadly, but that's out of scope here.

My experiences with running Insight checks this way have been that they help situate players in the game, even if it doesn't shed much light on the plot. They want more information, and if an Insight check doesn't provide it they either have to hope for the best and stay wary, or they try to verify claims in other ways (like investigating the claim after the conversation). The opportunity for NPCs to lie to or otherwise deceive the PCs is a part of the adventure, distinct from the players' dependence on me, the GM, to provide the information necessary for the players play the game at all.


At my own table I always say that someone is hard to read or that the player doesn't know on a failed check. The reason for this is because if a player rolled for example a natural one or other low roll that clearly failed and I lie, I've found it often causes issues with players feeling trapped by a decision.

For example: the player insights the injured man on the road and rolls a two. If I say "he seems genuinely in distress and honest" then the issue is this.

  1. my players are immediately alerted to the fact that this is not the case.
  2. they now feel that they cannot react to this knowledge without accusations of cheating or meta gaming.

In my experience this ends with players feeling forced into decisions they know are bad.

An inconclusive result in the other hand, gives no information, leaving your players free to interpret the situation any way they like. If they can't read whether the injured man is taking, they may decide to ignore or attack them, which could have consequences later or they may still spring the trap, but those in the party who are more mistrustful are free to be somewhat prepared for an ambush without fear of metagaming.

  • 10
    \$\begingroup\$ At tables that enjoy a bit of silliness the "obvious lie" can also work favorably due to your first point: When the party is approached by a shady guy trying to sell them a "magic item" in a dark alley and a character rolls a low insight it can lead to some fun roleplay to tell them that "to you, he seems really trustworthy". Despite the player realizing that he really isn't. \$\endgroup\$
    – Jave
    Commented Jan 14, 2021 at 8:26
  • \$\begingroup\$ @Jave I like that thinking because instead of the DM trying to decide what to say from a standpoint of the facts, the DM has to instead decide how the character interprets the situtation. In a sense, you tell the character what they think of the situtation which would reduce the important of true or false facts. \$\endgroup\$
    – DKNguyen
    Commented Jan 14, 2021 at 19:33
  • 1
    \$\begingroup\$ There are some rolls that players should not roll themselves, for exactly the reason mentioned in this answer. Especially if your table is prone to metagaming. In situations like that, it's best to roll for the player behind the DM screen and just give the answer based on that secret roll. The players won't know whether they succeeded or (critically) failed. They can guess, based on their skill values, but it keeps the suspense up. \$\endgroup\$
    – user440
    Commented Jan 15, 2021 at 10:14

Let me break this into two questions:

Is it ok to lie to the players ...

Yes, but ...

Your role as DM is nicely summarised in the How to Play section:

  1. The DM describes the environment.

  2. The players describe what they want to do.

  3. The DM narrates the results of the adventurers’ actions.

Since you are the one and only window the players have into the game, you should endeavour to make that window as transparent as possible. The players can only interact with the environment through their characters and those characters are not omniscient (usually), so it's ok if your description of the environment incorporates the limits of the characters' knowledge. However, it's not ok to deliberately misrepresent things that are or should be within the scope of the characters' understanding.

In summary, it's ok if the environment lies - it's not ok if the DM lies about the environment.

... rolling an insight

Ok, now we are talking about action resolution - the quasi-step in between:

  1. The players describe what they want to do.

  2. The DM narrates the results of the adventurers’ actions.

So, let's take a look at how we got here: what did "[t]he players describe [as] what they want to do"?

If it was "I want to know if he's lying" well then the appropriate response is "Well, I want more money and less body mass. How are you going to get what you want?" The thing about action resolution is that the player must actually state an action that can be resolved, not just state a wish about how they would like things to be.

The Angry GM says (and I agree) than an action needs both an intent ("I want ...") and an approach ("... by doing ..."). This gives you, the DM, the two pieces of information you need to resolve the action: the results if the player succeeds and the knowledge to assess if the action will succeed (no die roll needed), won't succeed (no die roll needed) or might succeed (die roll needed).

Let me illustrate how this works:

Player: "I want the orc dead [intent], by hitting it with my battle axe [approach]."

As a DM, you need to look at this to see if it will succeed (e.g. the orc is currently unconscious at the player's feet), won't succeed (e.g. the orc is 200 feet away behind a wall) or might succeed (e.g. normal melee combat). The resolution is either "the orc is dead" or "the orc is not dead" - the second has some nuances that the orc might be closer to being dead than it was before if the character hit and did some damage but not enough to kill the orc but those are mere details.

Finding the truth is no different from any other action the players might attempt:

Player: "I want to determine if he's lying or not by ...

  • ... looking for inconsistencies in his story."
  • ... questioning him about what he knows about the paladin and seeing if that matches what I know."
  • ... questioning the others and seeing if they tell the same story."
  • ... calling him a liar and taking out the tongue remover tongs and heating them in the fire."
  • ... buddying up to him to gain his confidence."

These all have the same outcome. If successful, the player knows whether the person is lying or telling the truth, if unsuccessful, they don't know or, at least, they don't know more than they knew before.

Success can be automatic, impossible or require an ability check. The approach determines what ability check to use - most are Wisdom (Insight) but I've also included approaches that use Charisma (Intimidation) and Charisma (Persuasion).

Remember we are resolving the action, the appropriate ability and proficiency come from HOW the players are doing it, not WHAT they are trying to do. Making the orc dead by hitting them with an axe uses a different ability than using a bow or a spell. Similarly, learning the truth can use a different ability depending on the approach.


The other answers cover two good mainstays ("hard to read" and blatant lies) well. I also suggest two more methods you can use with poor insight checks in D&D.

"They Seem to be Telling the Truth"

This carries slightly different shades of meaning from "they're hard to read." The player receives the same information (especially if you're consistent in using this phrase for both lies and truths over time). The character however is directed toward a different attitude.

I find using "hard to read" leads to characters being more suspicious since now this NPC is established as shady. If I use the more positive phrasing, the characters tend to treat the NPC as no less trustworthy than before, even if they retain their skepticism. My players often have the same attitude either way, but that slight nudge changes how the conversation and any subsequent information are received by the party, even if the players bemoan how they'll be betrayed the entire time. Since you can use the same phrase on a successful check (when it's true), they're never quite sure though.

Obviously, how the meta game plays out will differ table to table, but I've found this phrasing works across almost every group I've DMed for and it is currently my favorite.

Pan the Camera

The second method will not work in all groups, YMMV. I find that players who enjoy narrative tension and lean into role play usually love it while those who view the character as an avatar of themselves usually do not. It will not work if the players and the DM have an adversarial relationship. Also, if you use this, make it a recurring part of your game, not just the result of a single check.

You know that moment in a movie when the detective walks out of the room and the camera pans back for one last shot of the suspect's face, an item on the desk, or the locked wardrobe? Do that. Narrate something important the players missed in the conversation or scene. If they return, it's gone and the characters are none the wiser, but the players know they should work with you to find the missing information. Like many suspenseful movie moments, tension is built here with incomplete knowledge rather than complete ignorance.

In your example this might play out in a couple ways:

  • You mistake the cruel smile on the servants face to mean he is toying with you, feeding you lies to turn you against the paladin. You do not notice the small ring on his pinky finger that bears the paladin's crest.
  • Clearly the servant is lying. That is what he is known for after all... Player's leave scene Later that evening the local beggar enters the broken door and starts to scavenge the servant's room. The furniture is in disarray and almost everything of value is already gone but in the desk she finds two large books of genealogy and a small scroll sealed with the paladin's crest. "More fuel for the fire tonight!" she says with glee.

While I assume this does not fit with how you currently present the game, it can be a fun alternative that allows for memorable moments for players that draw a line between having fun and their characters' goals.


Two options here, that I have tried, and where successful from my point of view and the players.:

  1. Say "You don't know." In this approach, the players get exactly that - no sense of the intentions of this minion. This fits, I believe with the way checks work in 5e - if you fail an attack roll/athletics check you don't take damage/slide down a wall, but rather just fail at the attempted action (in this case gaining insight to their intention). This has worked with some groups, however other groups found this either not fulfilling, or not complex enough. For groups like this I prefer to lie (option 2)
  2. Lie, if, and only if, you know your players well - and they will not mind. With parties who have agreed, (in session 0) that I could lie, for their insight checks, I used either hidden rolls, or just trusted the players not to metagame.
  • 6
    \$\begingroup\$ How did these options work for you? Your experiences with these suggestions will be valuable to the querent, I think. \$\endgroup\$
    – Upper_Case
    Commented Jan 13, 2021 at 19:48
  • \$\begingroup\$ I'm happy to elaborate, but I cannot see what to expand -are you asking for actual examples of these in play? \$\endgroup\$ Commented Jan 13, 2021 at 19:53
  • 5
    \$\begingroup\$ Basically, yes. As currently written, this seems like a "try this" answer, which are discouraged here. Expanding on what worked well with each suggestion, or areas were they fell short, would be useful, as would expanding on why a DM might prefer one option over the other. A play-by-play example isn't necessary (though if you want to include one, you should feel free to do so). \$\endgroup\$
    – Upper_Case
    Commented Jan 13, 2021 at 19:56
  • \$\begingroup\$ that I have tried doesn't imply "and found they worked well". If you mean that, saying so would be a bare minimum; even better would be a description of what was good (although the "you don't know" approach is pretty standard and doesn't need much). If they didn't work well, describe the downside or problem so people know what to avoid. \$\endgroup\$ Commented Jan 16, 2021 at 22:16

Rolling or not?

We use ability checks if there is a chance for both success and failure. Here, the PCs want to ascertain if the NPC is telling the truth. When ascertaining information there is always the possiblity that you think a lie is obvious or that there is no reason whatsoever to assume a lie. This would mean that either the chance of success or failure is missing. This is a judgment call. Since your asking this question it seems you have judged that success and failure are possible. You could rule otherwise, but the question to ask yourself is not whether or not a deception occured but whether or not this is obvious to PCs.

Lying or not depends on the players and the game

This is mostly a question of preference. Whether or not this is okay depends on the players and the game you all want to play.

If you tell your players that you might lie to them on bad checks concerning gathering / checking information, and they have no objections, then it is okay. However, for this to work, you would need to use hidden checks (you roll for the players and don't reveal the result). If your players openly roll a 1, for example, and you tell them unrealistic things, they will instantly know and a very low roll is almost as good as a high roll. Obviously it depends on the question examined. If it is not a yes/no question, the lie will potentially not reveal a lot, except that it's wrong. If you use hidden rolls and the players get a crazy story, they have to make a judgment asking themselves if it's maybe true or if the roll was just bad.

If you feel bad about lying, your players have objections against it, or you do not want to use hidden checks, you should avoid lying. You just change the certainty and amount / relevance of the information. From "you have no idea what they're thinking" to "he's certainly lying / truthful".

Personally, I use hidden checks. For example:

  • Player: I want to check their body language to ascertain if they're being truthful.
  • Me: I'll make a DC 15 Wis (Insight) check.

If the check is successful, I tell them the truth. If it fails, I might not. Notice, that I tell the players the DC, so they can make a better judgment.

I also know GMs who use open checks without lying and it works equally well. It's really a matter of preference.


I'll take a different tack to most of the answers here and say this:

If your players prefer storytelling to "winning", lying only ever hurts the game

Even if the characters don't know the truth of the world, it can often be beneficial for the players to know. This all depends on what the players at your table do with metagame knowledge. If they metagame to "win" then keeping knowledge from them will help them actually roleplay their characters. If, on the other hand, they use metaknowledge to make the game interesting then they can use knowledge their characters don't have to facilitate that.

In the example you gave, by the time the players are interrogating the paladins follower they will already know the paladin is evil. When the characters can't get a read on the follower this gives them some options they otherwise might not have, such as specifically putting the paladin into a position of power that they might abuse later. This can help set up interesting scenarios.

I will admit, it can be hard to get the right table of players for this to work well, but if you can then it is often the most rewarding way to play the game.



Narrate a failed Insight checks by describe what the players are expecting or keep the status quo.

The exception is when it benefits your story or plans for what's coming ahead to NOT lie, then tell the truth (without letting them know its the truth) And listen to each players as they comment on whether the NPC is telling the truth or not... usually about 50/50 will say the opposite to each other. very funny.


You are constantly "lying" to your players, at all times. Why would this be different?

The characters don't exist, the world you're describing doesn't exist, etc. The PC doesn't exist, and the NPC doesn't exist. The core RPG experience is the interplay of pleasant fictions. You are describing the world their characters perceive, and that world is entirely fictional. If they perceive something that doesn't match the underlying fiction of the world, it's no more "lying" to describe that than it is to say that an elf is sitting in a tavern, let alone describing a room without automatically describing all the traps and secret doors.

So yeah: feel free to "lie" to them when they get an insight check.

On the more interesting question of how to lie to them, I've had pretty good success with the following approach:

Consider the insight check as the character's ability to see past the surface. The lower their check, the more facile their understanding is.

For example, if the scruffy, twitchy guy runs up and tries to get the PCs to go into a dark ally with him, a low insight check should say "this guy seems really skeezy". That's true whether the scruffy guy is a mugger or an angel who wants to give them a helpful artifact - sometimes things are what they seem to be, sometimes they aren't, and people with poor insight can only see the surface.

Deception then lets you control what you look like "on the surface." So usually, that means that if an NPC roles deception well and the PC rolls insight poorly, the result is probably "He seems totally trustworthy." If it's important to you to look untrustworthy, or shifty, or hungry, or whatever, you can do that too.

On your specific example, I would handle it by explaining that they can't rule it out, but that it doesn't match what they've seen, and this guy seems mostly interested in saying whatever thing will result in him not being immediately stabbed.

However, if you want to use answers like that, you have to use them all the time. This works great, but only if your answers don't have one mode for "bad guy" and a different mode for "good guy".

  • 8
    \$\begingroup\$ +1 for good answers in the second section. Your first section seems unhelpful -- that's a weird definition of "lying" and it's not one that most D&D players will share. \$\endgroup\$
    – Dan B
    Commented Jan 14, 2021 at 2:11
  • \$\begingroup\$ "how to lie to them" it might sometimes be useful to give multiple pieces of information. on a failed roll one of them (at least) would be untrue. The players know that, so they are left unsure in which one it is. Alternatively, you can reveal a different piece of information the players aren't after at the moment. If they roll to see if somebody is trying to mislead them, they might get that the NPC is concealing a northern accent. This information might be true but it's still a bad read of the situation. \$\endgroup\$
    – VLAZ
    Commented Jan 14, 2021 at 7:23
  • \$\begingroup\$ @DanB I added scare quotes and a little more explanation to make it more palatable. But fundamentally, the DM is narrating a characters'-eye view of a fictional scenario. If any description of the world can be called "lying", every description is a lie to a greater or lesser extent. \$\endgroup\$
    – fectin
    Commented Jan 15, 2021 at 0:29

There are lots of great answers here, but my opinion is that it's always ok to lie to players, as long as you recognize the impact it has on the players. If something is set at DC 15, and the players beat it after beating every single other roll they've made, changing the DC on the fly seems fine to me. On the other hand, if they've been having a hard time, and they roll up a 14, why not give it to them?

On the other hand, if someone crits on an insight, and you don't give them any information (and you actually have information), they will be pretty disappointed.

(Of course, the better situation is to modify DCs as you go depending on how things are going, just to be clear about that. But post-roll assessments are reasonable, in my mind.)

D&D Should not be balanced; it should be unfair, and it should be and fun for everyone (like life!).


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .