What The Rulebook Didn’t Include: Gumshoe (Part 1 of 2)

Note from the editor: This is a super deep, indeed comprehensive, dive by guest author Andrew Brehaut into making Gumshoe work. Part 2 will go up next week, and will pick up where this left off.

The Gumshoe family of games from Pelgrane Press has been around more than a decade now. These games, such as Trail of Cthulhu and Night’s Black Agents, excel at the kind of procedural narratives that are the backbone of episodic mystery fiction and TV. 

Gumshoe isn’t as trad as it might look at first glance. The system is frequently coupled with mysteries, a tough genre for many GMs to handle well when torn between creating an interesting investigation and time for meaningful characterization. I can help you avoid some of these gotchas! 

This is my approach to Gumshoe’s mechanics and how I work within the structural conventions. Most of my experience with the system is from Trail of Cthulhu and I will be using it for examples, but you should be able to generalize to the other games.

Before we look at the mechanics, a refresher on the mystery procedural genre. Unlike many popular roleplaying genres, procedurals tend to exhibit only a limited amount of character change or growth: they don’t get more powerful, and the protagonist’s dramatic arcs—if they exist!—are slow. Stories are typically episodic, with each episode bringing the characters into contact with a fresh mystery to resolve. This mystery forms the A-plot of the episode and is essentially the antagonist’s story: it is external—and often unrelated—to the protagonists. The protagonists’ stories are the B-plot and provide complications to the investigation and/or reflections of the antagonist’s story. Finally, the protagonists tend to be reactive, responding to the mystery first, and their own interests second.

Mechanical considerations

At its core, Gumshoe is two things: the conversation between players and GM (to lift PbtA language), and spending points from ability pools to change your circumstances. Whenever the player wants to do something where:

  • The outcome is in question,
  • Both success and failure are interesting,
  • And there is an ability in the game that governs that action

Then the GM offers the player an opportunity to spend points (for general abilities this coincides with a die roll, for investigative abilities it does not). If no governing ability exists, then it is entirely within the GM’s remit. The system is approaching diceless!

You may have noticed that Gumshoe characters have only abilities, no attributes. If you don’t have a suitable ability, the outcome is the GM’s to decide. Not having an ability isn’t the same as being unable to do it, it just means the player has no agency in the outcome. Shoot a gun without a single point of “shoot,” and you’re asking the GM to adjudicate.

From the players perspective the big mechanical question is always whether to gamble their limited resources (pool points) to gain what they want. There are some corollaries to this:

  • Pool points are valuable: every time the GM offers an opportunity to spend them to the player, their circumstances should change meaningfully in exchange.
  • Players must be transparent about their intent with a test, otherwise they risk wasting points. Likewise if the GM plays games with the outcome of the test, and therefore the players intent, they are abusing the players’ limited resources.
  • The results (success or failure) of a test should stand. Players cannot beg a retest without changing their circumstance, and GMs should not be frittering away pools with bogus tests just to trip a failure. Burning Wheel’s Let it Ride rule, essentially. This is supported by the rules about retrying failed tests.
  • Players can—and will—spend enough points to guarantee success with general abilities. This mirrors the way investigative abilities work.

Throughout the arc of an episode, the PCs get information and new resources that offer more choices, while expending their points reduces their options for how to act on those choices.

Difficulty numbers

Target values—die roll plus spend—are unfortunately referred to as difficulty numbers, but they don’t really reflect difficulty. Based on the rules text, I’ve developed my own heuristics: We already know that success and failure are both interesting, so if there is no obvious other consideration then I assign a target of 4 (e.g. a 50/50 chance of success). Pushing the number high or lower should correspond to how advantageous the potential narrative positioning and/or supplemental pools or refreshes gained from the test are. I will also adjust the difficulty by one point in either direction based on the player’s approach: needlessly risky or foolish is a one point penalty, a clever plan is a one point advantage.

What I’m looking at with these numbers is a range that produces some uncertainty with the player, but is not inherently unfair. I cap the target at 6, this way if you spend 3 points, you will generally get to, at worst, 50/50 odds.


The way investigative abilities work for clue finding falls out naturally from the way we think about tests with uninteresting failure cases: we don’t test abilities to deliver them. So it is with clues: they are untested. As long as you have rating (agency) in the ability and it is appropriate to use that ability, you get the clue.

My first guideline is to jettison the notion of point-spend clues completely. Early Gumshoe games, like Trail of Cthulhu, spent a lot of words on this concept and—I believe—it is fundamentally broken. Pelgrane seems to think so too, as they removed spending points on clues from later publications. It’s broken because every player has a large number of investigative points spread thinly across a lot of abilities. Ideally an individual scenario will not overuse one ability, and until the player has spent the points they don’t know what the clue is. This means that you don’t know what you are getting (so you cannot evaluate the exchange), and you have no way of knowing what future spend opportunities may be. This means the only reasonable choice is to just always spend the points when the GM offers a clue, which is mechanically unsatisfying. So get rid of point spend clues, and save those points for the much more interesting Special Benefits.

Another refinement that Pelgrane introduced in later games is the notion of free, non-core clues. These clues provide either color and/or some narrative advantage. Characters who have only uncovered core clues can make it to the final confrontation, but they will be ill prepared. Characters who have a broad range of clues will know about strengths and weaknesses, the antagonist’s plans, potentially gained NPC support, etc.

As GM, keep in mind that your perspective of the mystery is clear and simple. It might seem overly simplistic and not that interesting. For the players, they don’t have this view and are trying to interpret what they do have. Give small clues, and lots of them: this helps the players keep feeling like they are acquiring more knowledge and don’t need to treat the current set of clues they possess as a complete cryptic puzzle to unravel.

How small is a small clue, and how many is lots? My heuristic for determining how many clues, and how detailed they should be, is based on the investigative ability list: If the ability list (such as in Trail of Cthulhu, or Night’s Black Agents) is long with a lot of fine grained abilities, then I know I want to be delivering a lot of smaller clues. Alternatively, for games with short ability lists (for example The Yellow King) I’ll be aiming for fewer but larger (but not large) clues.

It’s very rare that I want to provide a big monolithic clue that contains a lot of concrete information the players need to decode. Instead, if I have something like that in mind, I make a dedicated prop the players can physically interact with, in the same vein as say Dracula Unredacted in Dracula Dossier, or The Armitage Files. Really, the prop exists as a jumping off place for improvisation and doesn’t provide anything concrete per se. Let the players dig through it, and anything the players find interesting becomes a real clue for the characters.

Published scenarios give a misleading impression of how clues are handled! Many of these scenarios give very crisply defined clue+ability pairs, but in practice the players infrequently try the exact approach in the text. If you are running published scenarios, think about alternative abilities that can surface the same clue as part of your prep. When devising clues for your own scenarios, worry more about what the clue is than how the players will find it; you can safely leave that to them.

Spending investigative points: Special benefits

When a player spends an investigative ability point (or a push in QuickShock Gumshoe or Gumshoe One-To-One games), their character gets some advantage. This can be some narrative advantage, general points (either as a refresh or as a dedicated pool of points), or both. The exact scope of these benefits varies from game to game so check the text of the game you are running. If the game is Trail of Cthulhu then you’ll want to check extra carefully as, due to a layout oddity, the big callout box describing example benefits appears on the page prior to the text explaining them. Worse, this page is on a different spread

I want to emphasize this point: as a player, you are aiming—in addition to uncovering the facts of the mystery—to build an advantageous position, narratively and mechanically, to aid you as your abilities become taxed. Leverage special benefits often. Gaining dedicated pool points are invaluable: the points are not tied to a specific ability so your options are more open. Also note that the text states In most cases, dedicated pool points stack on top of your rating. In other words, if you are particularly clever with your investigative spend you might get a pool of points that give you agency in an area outside the governing abilities of the game!

Next week: talking, fighting, squeezing goodness out of your NPCs, and how to build a mystery.

4 thoughts on “What The Rulebook Didn’t Include: Gumshoe (Part 1 of 2)”

  1. For the best way to present information in a Gumshoe game, look at the Trail of Cthulhu scenarios written by Graham Walmsley (on their own or collected in The Final Revelation).

  2. Thank you for the article! The latest GUMSHOE tech outside of QuickShock is in Esoterrorists 2.

Leave a Reply