How To Write Comedy Sketches

The Premise Is Everything

Most people start by trying to write a funny scene. That's backwards. You start with a single comic premise, which is basically a question: what if X was true in Y situation? "What if a couple broke up but their assistant had to facilitate it?" "What if a guy got promoted to manager at his own funeral home but his father still treats him like a child?" The premise carries the sketch. If it's weak, no amount of dialogue polish will save it. A strong premise gives you clear escalation rules. The audience should be able to predict, after seeing two beats, that the next one will be bigger. If they can't, your premise isn't structured properly. It might still be funny as a vignette, but it won't work as a five-minute sketch.

How To Write Comedy Sketches Without Losing Your Mind

I went through about forty drafts in my first year before I stopped trying to make sketches "clever" and started making them follow a logical but escalating pattern. The difference between a sketch that works and one that falls apart usually comes down to whether each beat raises the stakes or just repeats the same energy at a louder volume. Those aren't the same thing. Here's what I actually do now, in order. I don't always follow it perfectly but this sequence saves me more time than any other approach I've tried. Step one: write the premise as a single sentence. If you can't do it in one line, you don't have a sketch yet, you have a vibe. I'll write it on a sticky note and put it on the wall. If after a week I still can't articulate it in one sentence, I scrap the whole thing and start over.

Step two: establish the normal world in the first three lines. The audience needs to understand the baseline reality before you break it. Don't be subtle here. "Two accountants are reconciling a ledger" is fine. "Two accountants who are also secret werewolves reconciling a ledger during a full moon" is not fine at the top. You're building a foundation, not a fireworks show. The funnier the disruption, the plainer the setup needs to be. Step three: write the first escalation. This is where most beginners fail. They hit the same joke three times. Escalation means the situation changes, not just the intensity. In my sketch "Annual Performance Review" where an employee is being reviewed by a panel of three ghosts from different centuries, the escalation wasn't that the ghosts got angrier. It was that each ghost introduced a completely new historical standard of performance that made the previous one irrelevant. The ground moved under the character. Step four: find the turn. This is the moment where the character tries to resolve the situation and makes it worse. It's not a punchline. It's a structural beat. Think of it as the sketch inhaling before the final push. Without a turn, your sketch just climbs a hill and then ends. With a turn, it climbs a hill, falls back down, and then scrambles up the other side.

Step five: write the button. The button is the final image or line that closes the sketch. It should feel inevitable in retrospect but surprising in the moment. It's not a callback. It's the last thing the audience hears before the lights come up. A good button makes the audience groan-laugh, not just laugh and move on. I wrote a sketch called "Parent Teacher Conference" where the premise was that a kid's teachers were all animals playing the role of teachers. The first draft ended with the mother saying "I'll do my best" and walking out. The second draft had a button where the mother, after hearing every teacher's concerns, calmly says "Well, at least he's not failing any subjects" and then the camera pans to reveal her kid is also an animal. That turned the whole sketch inside out in three seconds.

Structure Patterns That Actually Work

Sketches follow a few recurring structures. You don't need to know the formal names. You need to recognize them so you can use them or subvert them. The game of the scene. This is the core concept from The Second City training model. The "game" is the repetitive comic element that makes the premise funny. Every good sketch has one game. Your job is to find it, isolate it, and then play it three to five times with increasing pressure. If you can't identify the game in your first draft, you don't have a game yet. Go back to the premise. Rule of three. This is the most basic pattern and also the most abused. Setup, setup, twist. The third item breaks the pattern established by the first two. The problem is that everyone knows this pattern, so using it predictably is boring. The solution is to make the third item genuinely unexpected while still feeling correct. Not "the third teacher is a lion" when the first two are a cat and a dog. That's pattern-matching, not pattern-breaking. Try "the third teacher is the school principal who is also a golden retriever who can only communicate through interpretive dance." Still rule of three, but now there's texture.

The absurd escalation. Start normal. End absolutely insane. The trick is that every step in between has to feel like a natural consequence of the previous step. If you jump from "the waiter brings the wrong order" to "the waiter reveals he's the protagonist's long-lost father who faked his own death," the audience will reject it. The jump is too big. You need the steps in between: the waiter refuses to accept the complaint, the waiter becomes increasingly personal in his criticism, the waiter pulls out a photo, the waiter reveals the photo is of himself and your parents at a beach in 1997. The callback structure. You introduce a seemingly minor detail in the first beat, and that detail becomes the engine of the final beat. This is higher risk because it requires the audience to track information across the sketch. But when it works, it feels incredibly satisfying. My sketch "IT Support" opened with the tech support guy mentioning casually that he'd fixed this exact problem before at a different company. Four minutes later, the caller turns out to be from that different company, and the tech support guy has to switch personas mid-call. The callback wasn't a gag. It was the entire plot twist.

Common Mistakes I See Repeatedly

Sketches that are too long. A five to seven page sketch is standard for most venues. Eight to ten works if the premise is particularly rich. Anything over ten pages almost always needs to be cut, not expanded. I once spent three weeks on a twelve-page sketch about a dystopian grocery store. When I performed a trimmed six-page version, the response was significantly better. The longer version had two premises fighting for attention and a midpoint that sagged. Cutting it in half didn't remove the comedy. It removed the fat. Multiple premises in one sketch. Each sketch should have one game. If you find yourself writing "and then also this other thing happens," you have two sketches. I wasted about six months in my early years trying to combine a sketch about a man whose shadow develops opinions with a sketch about a couple arguing over thermostat settings. The combined result was neither funny nor coherent. I split them six months later and both worked fine on their own. Writing dialogue instead of writing behavior. Comedy sketches are about what people do, not what they say. A character who says "I'm really upset about this" in a normal office setting is not funny. A character who has replaced every office plant with photos of their former boss and arranged them in a mocking smile formation is funny. Show the behavior. Let the dialogue be the byproduct of the behavior, not the source of the humor.

Ending too early or too late. Too early means you set up a premise and resolve it before the audience has had time to invest. Too late means you keep going after the humor has burned out and the audience is waiting for it to end. The test is simple: perform it for someone who doesn't know the premise. If they laugh at a beat and then immediately say "so does it end here?" you've gone too far. If they're still looking confused after your button, you haven't given them enough.

The Practical Reality of Writing Sketches

Sketch writing is not a linear process. You will write drafts that die. You will have premises that seem brilliant on paper and collapse under the weight of being performed. This is normal. The writers I respect most are the ones who treated failure as data, not as evidence that they should stop. My actual workflow: I spend about two to three hours on a first draft of a seven-page sketch. That's including the time I sit and stare at the wall trying to figure out why the second act isn't working. Then I step away for at least a day. When I come back, I can usually see the problems immediately. I spend another hour or two revising. Then I read it aloud. If a line makes me stumble when I'm reading it, it will make the audience stumble when they're listening to it. I rewrite those lines. Then I give it to one trusted person to read. Their feedback is worth more than any craft book because they're experiencing the sketch the way an audience will, not the way I intended it. I've found that writing with a partner changes the dynamic significantly. You catch each other's blind spots faster. But it also means compromise, and compromise is where some sketches lose their edge. I prefer to write alone and then bring in a reader afterward. The sketch stays cleaner that way.

When a Sketch Just Won't Work

Sometimes a premise is simply not sustainable for five minutes. You can have a genuinely funny idea that only works as a thirty-second bit. That doesn't mean it's a bad idea. It means it's a micro-sketch, a monologue bit, or material for a longer piece. Don't force a three-minute premise into a five-minute structure. You'll spend two hours padding it and the result will feel hollow. The reverse is also true. Some premises are too big for five minutes. A sketch about the entire history of the Roman Empire through the lens of a homeschooling parent is a great idea for a special. It's not a sketch. Recognizing the scope of your premise upfront saves a lot of wasted effort. I keep a running document of premises that didn't work. Most of them died within a page or two. The ones that survived past the first draft usually just needed a different structural approach. One of my premises about a man who could only speak in movie quotes was rejected in three different drafts because every version leaned on the quotes being funny instead of the situation being funny. The working version came when I stopped making the quotes the joke and started making the fact that his wife knew all the references and was using them against him in arguments. The quotes became texture. The relationship became the comedy.

Resources That Actually Help

Trust Exercise by Susan Bennett and Brian Myers is the most practical book on the subject. It's not theoretical. It's exercises, done in sequence, that teach you how to find and develop a game. I went through it cover to cover and it cut my drafting time roughly in half. Step by Step to Stand-Up Comedy by Greg Dean teaches premise development in a way that applies directly to sketch writing. The concept of "premises that generate premises" is exactly how you avoid the expansion problem I mentioned earlier. The Upright Citizens Brigade writing books are fine for absolute beginners but they repeat themselves. You get the same advice in three different formats. If you have the patience, read them. If you don't, read Trust Exercise instead and skip ahead to the advanced sections.

Watching sketches and breaking them down on paper is more useful than most people expect. Take a six-minute sketch from Mr. Show or In the Loop or Some One and write out the beat-by-beat structure. Note where the escalation happens, where the turn is, where the button lands. You'll see the same patterns repeated across completely different premises. That's the architecture. The content is decoration.

The Honest Limitations

Sketch comedy has a very high failure rate. Most sketches you write will not be good. Most of the ones that seem good won't work when performed. A professional sketch writer might have three or four usable sketches in a batch of fifteen. That's the industry standard. It's not a sign of lack of talent. It's a sign that the form is brutally selective. There's no workaround for that. You can't optimize your way out of it. You can only increase your volume and learn from each failure. The writers who last are the ones who keep writing after their fourth or fifth sketch bombs. If you're writing sketches for a specific venue, learn their format. A sketch written for Chappelle's Show has different requirements than one written for Kroll Show or Broad City. The structure, the tone, the acceptable length, the relationship between characters and audience. Ignoring these details won't kill your comedy but it will kill your chances of getting produced. I learned this the hard way after submitting a six-page political satire to a venue that specialized in character-driven absurdism. The sketch was fine. It was just in the wrong place.