I feel like I'm particularly bad at any kind of scene where I want to drop clues. I'm hesitant to use the term "crime scene" because it's not always being investigated by "police", and some of these times there isn't any kind of typical evidence.

I'm not even 100% sure that my problem is just the scene, but rather creating enough evidence to begin with. Other aspects I struggle with is witnesses - both witnesses with knowledge, and how to reveal it. Having useless witnesses for flavor, etc. I keep feeling that in general I give too little in these scenes, and everything I give is important.

Since I suspect this is a rather broad problem, I'd like to know if there are any Role Playing resources (sections of books, site, etc) that are specifically geared at teaching this portion of RPG storytelling?

I'm currently playing in the "new" World of Darkness 2.0, but I want answers on this not tied to the game system's rules.

  • 3
    \$\begingroup\$ This is not an answer, but I recommend the excellent GURPS Mysteries for some solid advice on the challenges RPG mysteries present and ways to mitigate and / or overcome them. \$\endgroup\$
    – gomad
    Commented Nov 10, 2014 at 13:35
  • 2
    \$\begingroup\$ @doppelgreener if I had wanted it more specific I would have made it so, this seems to be a perfectly fine question as system-agnostic the detail was given to clarify setting only. \$\endgroup\$ Commented Nov 11, 2014 at 2:03
  • 1
    \$\begingroup\$ @doppelgreener what does the tag wiki say? "You want solutions to the question that are not directly tied to the game's mechanics". I'm getting pretty tired of people editing tags in a way that changes the meaning of the question, and invalidates existing answers without OP input. \$\endgroup\$ Commented Nov 12, 2014 at 0:24
  • 1
    \$\begingroup\$ @doppelgreener I would suggest that removing system-agnostic would mean that I only want answers relevant to nwod whether or not adding nwod whilst still having system-agnostic means that... well, arguably the system-agnostic tag may not be a very good tag to begin with. \$\endgroup\$ Commented Nov 12, 2014 at 0:34
  • 1
    \$\begingroup\$ @doppelgreener I'm more interested in the role play aspect, than the roll play aspect ;) I don't think that always changes from system to system, but certain systems are probably not appropriate. \$\endgroup\$ Commented Nov 12, 2014 at 0:42

6 Answers 6


The big thing is how a game structures its facts when you want to have clue scenes.

Pre-established Facts

The GM decides what happened beforehand, and now the point is to have all clues and witnesses eventually point towards that fact. This is the most common way games handle things, but there is rarely good advice towards doing it well.

Start with Free Clues

Be sure to give players several free clues or facts in any scene they arrive in. These should require no skill or training whatsoever. You may choose to give more to characters who are trained in certain skills. Free clues require no roll, do not require players to say specific actions ("Oh, you didn't SAY you were looking under the carpet..."), but assume the character is competent and comes across these things. The free clues should not be false; they should be accurate, if incomplete.

Additional clues are given for appropriate searching, questions, and skill rolls as appropriate.

Witnesses are Forthcoming

Witnesses generally WANT the PCs to know something. They may tell everything they saw, but what they saw was incomplete. They may tell background information (i.e. "Those two were always arguing." "I think they had money problems."). They may tell part of the truth to advance their side of an issue or to hide some wrongdoing or shameful fact on their own part.

If a witness is lying or hiding something, you are best off telling the players, "It seems as though you're not getting the whole truth here. You're pretty sure this is generally accurate...but something is off." The players can later figure out how they're going to pressure the witness or verify facts on their own outside of this.

This is not because the PCs are mind-readers, but more importantly because the players are not mind-readers. The players should trust you, as a GM or narrator is giving them generally accurate information. This ALSO helps point the players in the right directions at where to look for clues or how solid to take anything they get. (Dogs in the Vineyard is the game that suggested this, and I recommend it everywhere now.)

Assume Competent Characters and Insider Information

If a character has a criminal background? Start giving them ideas based on their experience and the way they'd look at the situation. The character with a really high driving skill probably knows a lot about cars in general. The noble blooded investigator can identify specially crafted luxury items, etc.

Tailor the description to the characters. The information is not just about objects and history, but also about social scenes around these things as well.

"The scratches in the wall are weird, jagged. You look at the floor around it and find a bent nail, caked with a bit of plaster... and blood. Someone used a spiked bat, a cheap improvised weapon, definitely someone planning to hurt someone. This was a messy job, not a professional for sure. You can rule out the Espinozas - they wouldn't have been this stupid."

Motivations are Key

Figure out the motivations for the people involved. That's the best selling point. Most mysteries are built on greed, fear, jealousy, animosity and resentment. Notice how different this is than the usual rpg trope of "They were secretly trying to do something to do something with magic." When you do this, and put the motivations at the center of things, it becomes an exploration of who the NPCs are, rather than just "event X happened at Y location by Z person".

Dogs in the Vineyard lays out its "Towns" this way - you figure out that someone was a) breaking the traditions, and b) someone was being mistreated (sometimes it's the same person, sometimes not)...and just as often c) there was a reaction that may have been violent, extreme, or dysfunctional.

Sorcerer (in the Sorcerer & Soul supplement) uses specifically built relationship maps - tying together anyone who is related by blood, anyone who is romantically involved, then looks at the emotional ties/motivations above and beyond that. Who is doing something wrong, or hurting another, and what comes out of it, based on the tropes of detective noir mysteries. Often the criminal act ends up being very secondary to the real dirt going on between the people.

The Pre-planned Scenes Pitfall

Although you can have pre-planned events that the players are investigating, pre-planned scenes don't work as well. Many games attempt to create a linear or branching set of scenes the clues are supposed to point players towards, but this falls apart for several reasons:

1) Players may get on the wrong idea, and you end up having to railroad them into a correct path of investigation to get to another scene.

2) Players may use abilities or simply correct deduction and leapfrog several clue scenes, in which case, you have to railroad them into not solving things ahead of the planned scene build.

3) "Clue to clue to clue" is not actually a great narrative device. It shows up in action and pulp stories, mostly because the point of those stories isn't a good mystery or investigation, but simply excuses to show off a lot of locations and action. It really feels pointless, like getting strung along bad side-quests in a video game.

Games that allow you to improvise scenes work better. Players can come up with all kinds of methods of investigation: flipping a witness or accessory to a crime to helping them, finding unconventional ways to access information or follow clues, and of course, ways to solve a problem as well. And you're not stuck trying to find reasons that these don't work or waste time trying to push players back onto track.

Improvised Facts

There's several games which allow players to narrate facts on the fly, or allow the GM to simply create facts in the moment instead of pre-establishing them. Often these games tie the fact making into the mechanics, which makes the pacing and presentation a non-issue, and mostly the difficulty at that point becomes making sure the group ties together a coherent set of clues and doesn't splinter into people pushing the facts into different directions that makes a mess.

In these games, the GM's role is changed; you're following the outcomes the players are giving you instead of trying to find ways to throw clues at them. Since everyone is creating and improvising the facts together, no one knows the real outcome or answer to the mystery until it is solved which makes it pretty exciting for everyone involved.

Inspectres is a game that builds it's investigation completely on this and might be worth checking out as an example.

  • 2
    \$\begingroup\$ The middle ground between making up facts on the fly and having a scripted set of clues to find is big. I've often used a technique of deciding what action took place at the crime scene and developing a strong idea of how that looked, and how it may have left evidence behind. The clues aren't all clear enumerated, but I have a cluster of ideas. I'll detail out and narrate whatever evidence makes sense based upon the PCs' actions. These "clue options" are semi-redundant, so the PCs are sure to progress no matter what. \$\endgroup\$
    – Jessa
    Commented Nov 9, 2014 at 20:24
  • 2
    \$\begingroup\$ There's a reason I look to pre-established facts rather than pre-established clues. Facts you can create a lot of clues based on, but what I mean is that who did X and what happened, is established rather than open to creation in play. \$\endgroup\$
    – user9935
    Commented Nov 9, 2014 at 21:04

The GUMSHOE system (used in Trail of Cthulhu and some other games) is a whole game system designed to keep things going during investigation games.

Investigation is generally hard to play in some systems because either you flat out give the information to the players with no effort on their side or there's some kind of roll that, if failed, stops the PCs on their tracks.

The GUMSHOE system automatically gives players the needed hints, but adds insights on a successful roll and lets the players draw the consequences of their discoveries.

Some other good investigation games procedures focus too much on letting the players decide the details (Tremulus, Crime Scene Unit), which is not compatible with the usual, pre-planned plot (as loose as it may be) of WoD games.

Lastly but not leastly, once a novice WoD ST asked for some advice on playing his campaign to our WoD ST and some precious wisdom got in my hearing range. It's not spot-on on investigation but it's closely related IMHO.

Always have your "crime scenes" reveal one truth the characters were looking for, but add one or two new questions. Drop full solutions of something the characters wanted to find out but show some new elements that will click togheter later.

  • 2
    \$\begingroup\$ so would reading up on the GUMSHOE system be helpful? (sounds at a glance like an extended roll where you give more info for more successes) \$\endgroup\$ Commented Nov 9, 2014 at 17:57
  • 5
    \$\begingroup\$ Reading up on GUMSHOE would be useful for point of view. The crucial idea is finding the clue is boring, what's exciting is what they do with it. \$\endgroup\$
    – gomad
    Commented Nov 9, 2014 at 21:52
  • \$\begingroup\$ Gumshoe would be well worth a look. There's a lot of excellent guidance for exactly this kind of thing in it \$\endgroup\$
    – Wibbs
    Commented Nov 10, 2014 at 7:42
  • \$\begingroup\$ "the usual, pre-planned plot of WoD games" isn't useful, and hasn't been true for about fifteen years. \$\endgroup\$
    – Jim Kiley
    Commented Nov 11, 2014 at 21:08
  • 1
    \$\begingroup\$ @JimKiley care to expand with some reference? My ST is still doing it that way (and his games keep being great, so it's not something I'm dissing) \$\endgroup\$
    – Zachiel
    Commented Nov 11, 2014 at 21:48

I watched a whole lot of Law & Order with an eye toward trying to understand how they structure their mysteries. I noticed a few things that I think ought to help run a mystery game. (Life has gotten in the way of testing my theories in the game I was to run, so do report back if any of this helps!)

First, a mystery is not a confusing story--it's a simple story about a perpetrator, a victim, and a crime scene, that the characters learn out of sequence. Supposing we're talking about a murder, there is a victim, and there is physical evidence. The detectives will have physical evidence to check out, and they'll want to learn about the social network that the victim was a part of. The clues and interviewees lead to more places to look for more evidence.

Bringing me to the second point: the detectives spend as much time or more ruling possibilities out than learning about them. The common problem GMs face is not wanting to give out too much information; I suspect that the thing to do is give out a good amount of information that looks suspicious, so that the characters turn up new evidence while ruling out suspects.

The elements of a mystery game, then:

  • nameless NPCs who are forthcoming with information, though they don't have any idea how it fits into an investigation;

  • named NPCs who act suspiciously and lie to the PCs, but because they are concealing a secret that doesn't directly relate to the crime;

  • a perpetrator who seems like they can be ruled out, but whose guilt will be revealed when facts that come up while the PCs are ruling things out show the motive, means, and opportunity of this perpetrator. Justice is served, etc. etc.

Like I said, these are theories that haven't been borne out by play. But I stand by this idea that a crime is a simple linear incident, and an investigation is a nonlinear process of alternately expanding the network of physical and social links, and ruling things out, until only the perpetrator remains.

  • \$\begingroup\$ These are some really good points. I recall one frequently used device in police procedurals is the innocent bystander who doesn't want their spouse to know they were cheating on them, so they lie about where they were ("no, I wasn't at that bar last week", or "no, I wasn't close enough to Mrs. Jones yesterday to see if she was acting unusually"). Children lie to stay out of trouble ("No, I wasn't under the bleachers during the game", but they are lying so they don't have to admit they were smoking rather than lying to protect whoever killed the new kid and left his body under bleacher #3). \$\endgroup\$ Commented Apr 19, 2022 at 10:39

"Just the facts" just isn't enough

Plan more flavor and context into your scenes, and your story. WoD games often lend themselves to strong thematic overtones, so don't skimp. Small side stories, or a growing tapestry of NPCs can both add a feeling of depth to the play experience. As you write later scene, call back to those "useless" or "flavor" elements from earlier.

These threads don't support the factual investigation story, but they support the emotional investment the players should develop. Recurring NPCs, even if they are unimportant people, are characters the player can start to get attached to, or develop other feelings towards like pity, anger, or curiosity. Less "important" people can also illustrate the human impact of what the antagonist(s) and PCs are doing in the world.

Add enough grit and detail to create a solid emotional journey, and the story will feel more meaty.

More facts don't hurt, either

That said, writing a good story/mystery can be ten times harder for most of us than telling it back during play. I've fallen into the trap of carefully coming up with some scheme for the players to learn about, thinking it was great because I spent so long on it and put so much effort into it. Then, they quickly gobbled it up and were ready for more... that I didn't have!

Try just adding more facets to the story and clue details to found.

  • \$\begingroup\$ Most of this answer doesn't seem to actually address the question \$\endgroup\$
    – Wibbs
    Commented Nov 9, 2014 at 21:40
  • \$\begingroup\$ The question includes concerns about using enough evidence, having "useless" witnesses for flavor, and how to make better sue of witnesses. This answer is attempting to address those concerns. \$\endgroup\$
    – Jessa
    Commented Nov 9, 2014 at 22:29

With questions like these, I always point to Alexandrian's Three Clue Rule essay.


As Alexandrian says, have at least three clues per thing that you want the players to figure out. And as Bankuei says, make at least some of them obvious. Generally I use two tiers when making a mystery. There are three first tier clues that point to the person/location/thing that the players need to run into. But when the players enter the mystery, what they find first are second tier clues to the first tier clues or places they can find other second tier clues.

So lets say we have a murder mystery. We come up with three first tier clues: the guy who acted as lookout while the murder was committed, the murder weapon, and the fact that the two victims knew the murderer.

We decide that we have three second tier sources of clues: The two crime scenes, and the CCTV network that one player has access to.

At the first crime scene are stab wounds that will match the murder weapon, a witness who saw the lookout and can describe him, and the matchbooks that the lookout dropped at his lookout point, which points to his favorite bar.

At the second crime scene are stab wounds that will match the murder weapon, a witness who saw where the murderer disposed of the murder weapon, and a series of emails between the murdered and murderer which shows they had business ties.

The CCTV network will show the outlook at the second crime scene, and the fact that the murderer frequently visited both victims. He looked very angry when he visited the second victim before the day before the murder.

To shore up the outlook clues, we'll also decide that the outlook is a petty crook, so his name and address can be found by finding him in the criminal database (if they know what he looks like).

Finally our first tier clues: The murder weapon has the murderer's finger prints and his initials on the grip.

The outlook can be made to spill the beans. He knows exactly who the murderer is.

And digging into the emails and other records will reveal that the three were engaged in shady business deals, where the murderer was getting the short end of the stick.


I can't post this as a comment, but I feel like it's worthy of an answer anyway. it expands on what Bankuei answered about pre-planned scenes.

Be flexible in your plot

If at all possible, allow the investigators to go at a faster pace than you accounted for. Maybe they figured out where the criminals will strike next before you expected them to, even before the criminals actually strike there. Maybe they decyphered what the big bad is up to at such an early time that they get there before he actually starts a crucial part of his master plan. In that case, avoid the temptation of slowing them down by throwing a couple of not-so-random encounters at them. Players like to think they outsmarted the GM, but don't like it if the GM punishes them for it.

Instead, allow them to halt the big bad before he even gets going. Or let them warn the town watch about the criminals, or even let them go after them themselves. Then, when they're patting themselves on the shoulder, PLOT TWIST! The party gets arrested by the town watch because the watch doesn't know that the esteemed noble is actually a demon. Or the criminals get bailed out and immediately attack the party at their inn.

Or the party is actually completely wrong in their clues. This is a goldmine. Go with the flow. Let the party think they caught a major criminal, while they actually caught a petty thief with no relation to the plot. They think that everything is fixed, but when you then unleash the next part of the plot, they actually realize that there is more to it all.


You must log in to answer this question.

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