A complete guide to scenario design in the convenor's own words, best suited to short scenarios — the source that Lectures 1, 2, 4, 6, 8 and 10 draw on directly.

The Convenor’s Scenario-Writing Guide

This is a complete guide, in the convenor’s own words, covering everything from picking an inspiration to writing the checks and NPCs that fill out a finished scenario. It’s best suited to a short scenario — a single throughline you can run in one or two sessions — rather than a sprawling campaign. Lectures 1, 2, 4, 6, 8 and 10 draw directly on the sections below.

1. Inspiration and genre

Decide what genre the scenario is in before anything else — you can’t write a story without the inspiration for it. Only afterward should you start pulling in the Mythos’ generic elements, checking which of them actually fit that inspiration, and folding them in to tell a Cthulhu-flavoured story.

There are two broad categories: a straight Call of Cthulhu scenario, or a fan-fusion scenario that borrows Call of Cthulhu’s rules for an entirely different setting — for example, a Lord of the Mysteries fan scenario.

Finding inspiration. Look for something you’re already interested in — it only needs to be interesting enough to carry a complete story. That might be a specific Mythos entity or Great Old One you like, or an original creature of your own. For example: Ithaqua’s range is defined as within the Arctic Circle, so a scenario built around it could be set in the northwestern Canadian wilderness in late autumn or winter — the setting of a Derleth story about a Miskatonic expedition attacked in that region, after which a female doctor is abducted.

Or find inspiration in your own daily life, then choose Mythos content that fits it afterward: something you saw in the news, a cold case that’s been open for years, a summer camp that went wrong, a scientific theory or experiment you got curious about, or a strange trinket you happened to see — say, a piece of jewellery in a shop window that you can’t stop thinking about after you get home. That’s usable.

Generic Mythos elements, at the level of the setting itself: ancient, unknown, chaos, and space.

A worked example. The “Dulce Base” incident is a well-known urban legend about an alleged 1979 firefight between Americans and aliens at an underground facility. From a story like this you can extract concrete, reusable pieces: the military base itself; underground “Greys”; laser weapons; a finger vaporised instantly; alien experimentation on humans; and unnatural underground tunnels.

Fusing the pieces. From that Dulce Base backdrop, you could connect to the K’n-yan from Trail of Cthulhu — an underground empire beneath Oklahoma lit by blue radiance, home to bizarre creatures; long-skulled, brown-skinned people who fled the surface after a cataclysm. They mainly worship Tulu (i.e. Cthulhu) and Yig; they once worshipped Tsathoggua too, but abandoned and destroyed all its shrines after finding something in the dark abyss of N’kai. K’n-yan communicate telepathically (some gifted individuals can even create mental illusions) and can dematerialise, existing simultaneously on the material and astral planes — able to pass through solid matter, or dematerialise objects they touch. Their weapon is a disintegrator ray-gun whose damage can’t be healed; their protection is an advanced “spirit-cloth” effective against any attack.

2. Planning the scenario’s flow

A useful frame for the planning stage: time, place, characters, cause (the scenario’s hook), development (the scenario’s body), and result (the scenario’s ending).

Era. For example, the 1920s — where weapon and technology choices should match the decade, so it’s worth knowing the period’s communication and transport options and its weapon models — or a modern-day setting instead.

Duration. If the plan is a single underground-tunnel expedition, keep it to 1–3 in-game days.

Place. A scenario set in an underground cave or a military base doesn’t need to fuss over which country or city it’s in. For example: a secret US military base near Arkham, Massachusetts, or a newly discovered archaeological site built around an unnatural underground passage. A military-base setting limits the roles and backgrounds available to players — base staff, soldiers, medics. An archaeological-dig setting can instead be framed as a Miskatonic University expedition — which fits neatly, since Miskatonic’s active period in the setting is mostly the 1920s–30s.

Hooks. Players can be pulled into the scenario passively — everyone just rented a cabin in the woods for a memorable Christmas, and only after arriving do they discover the forest has a Wendigo in it; or they’re simply eating at a fast-food place when they’re suddenly saddled with guardianship of an orphan; or someone wakes up to find they’ve lost their memory. Or players can choose to get involved themselves — they receive an anonymous letter out of nowhere, or a friend or family member goes missing, or they receive a death threat; or the police, stuck on a difficult case, ask civilians for help, and the PCs decide to form a team.

Worked example: Miskatonic’s archaeology team loses contact shortly after entering an underground cave, and three days pass with no one emerging, so the university organises a search-and-rescue team to investigate. The missing-persons angle creates more tension than a plain archaeological expedition — and the human-experimentation angle noted above is good material to fold in as well.

Structure: open vs. linear.

Worked example (linear): players take a lift down into an underground tunnel → explore the tunnel → discover the K’n-yan research base, or come under attack → the players are killed, used as test subjects, or escape with their lives. The core scenes here are only the tunnel and the K’n-yan base — but two scenes is too short to carry enough clues, so add scenes purely to deliver clues: a pre-tunnel investigation phase, talking with the people organising the rescue effort to get background from the university, and exploring the archaeology team’s surface camp — which can be extended into a nearby indigenous village near the tunnel, giving both social and literature-type clues. Aside from the K’n-yan (who can’t be talked to), this gives players chances to gain clues through conversation, so a player who put points into social skills actually gets to use them.

Design principle: remember the scenario is for players, so when designing scenes, clues, or events, always account for what players might expect — some players enjoy combat more, others enjoy gathering clues and reasoning out the truth.

Refined flow: explore the surface camp / talk to the rescuers → enter and explore the tunnel → come under attack / enter the base → ending. The tunnel-to-base transition is where you lean on the Mythos creature: use the K’n-yan to build atmosphere, which can also drain Sanity or set up the reveal and climax.

Worked example (the tunnel scene): the tunnel itself is monotonous, so clues are hard to place and exploration space is limited — but this is where the K’n-yan’s telepathy comes in. Have the K’n-yan notice the players shortly after they enter the tunnel, and use guided mental illusions to lure them in as test subjects. This enriches the tunnel section and builds atmosphere, since illusions are usually something that only happens to someone on the edge of losing their mind. Example: when the players try to backtrack, they find the spot where the lift used to be is simply gone. This pressures players’ Sanity while also stopping them from feeling like they can just give up halfway. As long as it doesn’t break the game’s rules, an existing setting can be extended and adapted quite a bit.

3. Clue design

Clues split into two broad categories: guiding clues, which give player characters direction for where to explore and act next, and revealing clues, which cover everything else — the solution to a puzzle, information about the scenario’s Mythos content — mainly used to push the plot forward and explain it, while also working as a small reward for a player character who actively investigates.

On guiding clues. A session is really a chain of scenes strung together, and it’s the guiding clues that link those scenes and locations. Neither the Keeper nor the players want a PC stuck in one scene not knowing what to do — it wastes time and saps everyone’s motivation to keep going. Even a player with big ideas who can decide their own next investigative target can’t be relied on to always pick the location that happens to trigger the intended plot; for a new Keeper especially, players wandering into an unplanned scene is close to guaranteed to stall the session, or force the Keeper to hand over a core clue early. These clues don’t take much deliberate thought — anything connected to your next intended scene, placed so it doesn’t feel out of place in the current one, works. For example: if you want the PCs to head to the library next, put a library card, or a book recently borrowed from that library, in the current scene, or have an NPC mention in conversation that “the missing person visited the library the day they disappeared.”

There’s also an implicit, semi-metagame convention in CoC: anything the Keeper specifically calls out in a description is assumed by the players to matter. For example: “In the victim’s suitcase you find a change of clothes, toiletries, and an oddly shaped little statue.” Any player, however low their character’s intelligence, will go look at that statue — because in real life, you wouldn’t miss it either.

Guiding clues can be explicit or implicit, and you can offer more than one direction. Treating “no guiding clues” as a test of the players’ deduction, or taking pride in stalling them, isn’t good design — unless your players tell you directly they don’t want guiding clues, their enjoyment comes first.

On revealing clues. Harder to generalise, since they depend entirely on the specific story — but for a scenario investigating a murder committed by a person or Mythos creature, the clues roughly split into three groups:

  1. Clues around the victim’s body — time of death, extent of damage, external examination findings (wounds), central-nervous-system damage (brain and spinal cord), internal examination findings, whether poisoning is present, and so on. Even if you don’t want PCs performing the autopsy themselves, provide at least a reasonably complete autopsy report — the manner of death is itself information about the killer or creature: where the attacks were aimed, its typical methods, and, in a serial case, even what it seems to select victims for.
  2. Clues around the crime scene — physical evidence such as fingerprints, footprints, claw marks, bullets, documents, and burned material, or biological evidence such as blood, saliva, hair, plant pollen, or a deep one’s scales. Guiding clues often live here too, but the crime scene is one of the best places to lay out detail and hint at the killer or creature. A scenario aiming for a strong Mythos atmosphere should use every means available to convey a creature’s strangeness and horror; even a scenario aiming for a “perfect murder” with no Mythos element at all should still make its misdirection something players have to think through, rather than “the killer was clever and cleaned up all the evidence, so the PCs find nothing” — the scenario still has to be playable.
  3. Clues about the plot itself — running a session is, at bottom, a way of telling a story, and a story without a plot is unbearable to sit through. Whatever genre or style you’re writing, you should already know what story you want to tell, and your clues exist to “explain” that story to the players in place of your own voice — a Keeper can’t spell everything out through narration without the session losing its point. If the story centres on a Mythos creature, pass along creature-related information through old records or a nursery rhyme passed down by word of mouth; if it’s a grudge-and-vengeance story, pass the plot along through tear-stained diary pages, or words scratched into a wall.

Delivery formats. Literature-type clues (old books, specific documents, diaries, notes), social-type clues (obtained through talking with an NPC), and trace-type clues (physical evidence). Splitting clues across these formats matters mainly because if PCs spend the whole session buried in the library reading one book after another, that gets strange fast — and too much repeated content is a flaw in both scenarios and novels.

4. Puzzle design

Puzzles broadly fall into a few types: object-hunting puzzles, pattern puzzles, wordplay puzzles, and decoding puzzles. None of these types ranks above another — which one fits depends on the clue’s content.

Difficulty isn’t the same as quality. A “hard” puzzle in a solving game isn’t automatically a “good” one. A puzzle exists to be solved, and a scenario isn’t ranked by its difficulty — only by whether the players got enjoyment from it. If you know your players well, set puzzles that match their taste and still take some real thought to crack, which both drives their motivation to solve and gives them a sense of achievement once they do. For a one-off group you don’t know well, either mix several puzzle types into the scenario and pick on the fly based on how players react, or be flexible enough to build an appropriately-difficult puzzle on the spot — though new Keepers or scenario authors attempting this without much practice risk hurting the experience, so an ad-hoc one-shot is often safer run with an already-tested scenario.

Purpose first. Be clear about why you’re setting a puzzle at all. If it’s purely to hand over a clue, why not just have the PC roll a Spot Hidden or Library Use check instead? For some situations, a skill check genuinely is the better call — it’s one of the distinct pleasures CoC can offer. But when you deliberately set a puzzle for a specific clue, you should have something you want to communicate to the players, and the puzzle and its solving process is exactly how you pass that idea along.

Worked example: (a) investigating a missing-person case, PCs use Spot Hidden and find a badge with a strange design in the desk drawer; (b) investigating the same case, PCs use Spot Hidden to notice a bookshelf against the wall looks like it can be pushed, and behind it is a hidden compartment holding a small four-digit combination safe, which the PCs open using a reference chart found under the bed, revealing the same badge with the strange design. Physically, is there any difference in outcome? None — both times a badge is found, and the second version might even burn ten or twenty extra minutes on lengthy thinking and description. But is that actually wasted time? Add one assumption: the badge is a cult member’s ID tag, and by examining its design PCs can trace the missing person’s connection to that cult, giving them their next investigative direction — in other words, this is a guiding clue. Given that, the ten or twenty minutes is completely worth it: a badge tossed carelessly in a drawer might not draw attention, but a badge locked away behind layers of security absolutely cannot be ignored. That’s the message this puzzle — not a hard one to solve — naturally communicates: this badge matters a great deal to the missing person, and is very likely a key clue to finding them. Mixed in instead with other information obtained through plain skill checks, PCs may well overlook its guiding role and stall out — especially likely in a linear scenario — and the cost then is far more than the ten or twenty minutes.

So: a puzzle is worth setting for clues that are clearly guiding and important to the plot — but given the time cost and the scenario’s length, keep puzzles limited to those key clues; everything else can be obtained through the faster route of a skill check.

Design points:

  1. A clear goal and a clear solution. This is the core of every puzzle — but for the scenario author, what matters more is being clear yourself about the solution, since PCs often won’t know in advance which clue a given puzzle points to (otherwise why bother solving it?) — see the badge example above.
  2. Gradual difficulty. You wouldn’t hand a child who’s still learning their ABCs an entire book in a language they can’t read and expect them to get through it. Make sure your puzzle guides players step by step from a shallow starting point toward the solution, especially for players used to just looking the answer up online.
  3. Don’t invert means and ends. Don’t design a puzzle purely for the sake of designing a puzzle — always remember a puzzle exists to serve the plot, not the other way around.
  4. Running puzzles in parallel often pays off. Give PCs two or more puzzles related to the same answer at once: when they’re stuck on one, they can work on the other, then use its result to work backward into the one they couldn’t crack. This asks more of the scenario author, but tends to land well at the table — and for a written scenario it isn’t hard to pull off: place multiple guiding clues in the same scene, and PCs get more varied options for where to investigate next. Look closely and you’ll notice this is, in effect, exactly what an open-structure scenario already is — an open scenario isn’t some especially advanced technique, it just asks a bit more of the Keeper’s ability to track the timeline and adapt on the fly.

5. Check and Sanity design

Don’t let a check become the point where the session gets stuck. In a multi-branch scenario you can use checks to gate how many clues PCs pick up, but in a simpler linear scenario, if a PC misses a key clue — or the clue pointing to the next scene — because a check failed, the Keeper is left either improvising a new thread outside the existing structure, or letting everyone walk toward a total party kill. For a linear scenario, think carefully about whether to gate key or guiding clues behind a check at all; setting a moderately difficult puzzle instead, for interest and controllability, is one reasonable option.

A check still needs narration, not just a pass/fail readout. Example: Keeper: “You open the cabinet to see what’s inside — roll a Spot Hidden.” Player A rolls and fails. Keeper: “You don’t notice anything worth mentioning.” Player B rolls and succeeds. Keeper: “You notice a revolver sitting in plain view in the cabinet.” Wait — a revolver isn’t worth mentioning? This is also why players joke “guess I’m blind” or “guess I’m deaf” after a failed Spot Hidden or Listen roll — and, mechanically, that’s actually correct; the only problem is in the wording. The fix is simple: change the success description to something like “you notice a hidden compartment in the bottom of the cabinet — one you happen to recognise from a book or film you’ve seen — and inside it is a revolver.” That explains why the other PC missed it, while still giving the successful roll a sense of achievement. This is something you can avoid entirely at the writing stage, just by checking the rulebook’s description of the relevant skill before setting a check — each skill’s entry already spells out, with examples, what kind of thing that skill is meant to reveal (in this case, that Spot Hidden should turn up things that are “not obvious”), and your success/failure text should reflect that.

Match how many checks you call for to your specific players’ taste. Different people want different things from a session — the satisfaction of a successful deduction, the thrill of a good fight, the tension of a hopeless situation — and checks are no different: some players love the uncertainty and surprise of the dice, including a spectacular failure; others can’t stand how uncontrollable dice are. Take that into account: for players who dislike checks, swap a clue’s delivery for a puzzle or a wordplay game instead; for players who dislike combat rounds, swap the scenario’s intended dramatic peak from a fight to a negotiation with the mastermind, or an unwinnable arrival of an Old One instead.

On Sanity checks. Madness is the result of a Sanity check, not its purpose. Tearing up a character sheet doesn’t mean “you won” or “the player just wasn’t as clever as you” — a Sanity check should only reflect the possible bad outcome of an action, or a situation the rulebook defines as one that may cause a Sanity check, nothing more. (If your players explicitly ask for more Sanity checks because they enjoy them, all of the above is void — give them a Sanity feast.)

When a Sanity check might cause temporary or indefinite insanity, consider dialling down the difficulty of the fight or negotiation that follows. Treat a PC with a strong chance of falling into temporary insanity as effectively absent for a while — at least for the immediately following combat or search — and set the following checks or combat difficulty a little lower than you would with everyone at full strength.

6. NPC design

Be clear about the NPC’s purpose. Before placing an NPC in your scenario, decide what role they’re actually meant to play — guiding PCs to the next scene, delivering a hidden clue, or acting as a pawn sent by the mastermind to mislead them — and write that down explicitly. This matters because it preserves your (or another Keeper’s) control over the scenario: a session’s actual events are hard to predict, so an NPC without a clear guiding purpose is exactly what leaves an improvisation-averse Keeper stuck, or pushes them into handing out clues too early.

Also worth noting: “lots of NPCs” doesn’t mean “rich content,” especially in a linear scenario. If your scenario is investigating a murder where the killer is an outsider who happened to pass the victim’s building and struck on impulse, you could give every resident in that building a full name and a few paragraphs of backstory — but how much of that would actually be useful? Dense, richly detailed scenarios are fine, but whatever you’re writing, the proportion of your text that’s actually load-bearing determines the quality of the work — so it’s worth starting from the parts that look simple but matter, rather than padding.

Give the NPC a distinctive trait. This might be an old point, but it’s still the real measure of whether an NPC’s characterisation landed. You might forget most of a scenario’s PC names, but remember an NPC who “looks down on people with low IQ,” or forget every PC’s nickname but remember the one NPC who mercilessly shot down the group’s signal without any mercy. That’s what a successful NPC looks like. In practice, even an experienced Keeper’s NPCs can start blurring together due to mood or ingrained social habits — that improves with practice and self-awareness — but it shouldn’t stop you from giving NPCs clearly distinct traits at the writing stage.

In play, that trait usually shows through appearance or personality, and dialogue should fit the NPC’s identity and era: it’s hard to believe a noblewoman rhapsodising about slum life, or a lifelong deep-sea fisherman who can rattle off the finer points of livestock postnatal care — that’s a mismatch of identity. If you’ve set the scenario in the 1920s, a physicist obviously can’t pull out a photo of a black hole to show you (unless a hound has already taken an interest in him) — that’s a mismatch of era. These details might seem minor, but they’re part of what makes a scenario feel a little more convincing.

Avoid long NPC-to-NPC dialogue. Running a session is a story you build together with your players — not a performance you put on for yourself while they watch. If you genuinely can’t avoid having two NPCs negotiate at length with the PCs only listening in, summarise the gist of that exchange for your players instead of reading it out line by line.