How to Do Stand-Up Comedy: A Beginner's Guide to Writing Jokes and Owning the Stage
Section 5 of 16

How to Structure a Joke: Premise, Punchline, Tags, and Callbacks

8 min listen Updated

A good premise is a gift that keeps giving. The comedian Jerry Seinfeld built an entire bit around the idea that the airline industry treats a delay like it's good news — "we're hoping to get you out as soon as possible." That's not one joke. That's the soil a dozen jokes can grow in. The frustration of being trapped, the fake cheer of the announcements, the seat itself, the food, the guy next to you. One angle, mined for everything it's worth.

That's the thing nobody tells beginners. They think a great comic is someone who has a hundred unrelated jokes. Actually, a great comic is someone who has a handful of strong premises and squeezes each one until it's dry. So before you ever worry about being funny, the real work is finding an angle worth standing on — and then learning the parts that let you build a whole structure on top of it. That's what this section is about: naming every part of a joke, and showing you how they stack.

Let's start at the foundation, because everything else sits on it. The premise is the underlying idea, the point of view, the thing you actually believe or noticed. The comedy site Mint Comedy puts it cleanly: the premise is the larger idea or point of view underneath a joke, and multiple jokes can share the same premise. Their example is dating apps — the premise might be "dating apps have made everyone worse at meeting people in real life." Notice that's not a joke. There's no punchline in it. It's a stance. And every individual joke you write under it works on its own, but they all sit on top of that one belief.

Here's why this matters more than any other single thing in joke-writing. Finding strong premises is half the battle — maybe more than half. A weak premise can't be saved by clever wording. A strong premise practically writes the jokes for you, because once you've located a real tension — something genuinely annoying, or absurd, or true that nobody says out loud — your brain starts generating angles automatically. This is the part most beginners skip. They reach for a punchline first and try to reverse-engineer a setup. Backwards. Start with the thing you can't stop thinking about, and the jokes will come.

So here's the test before we go on. If someone stopped you and asked what the difference is between a premise and a joke — what would you say? … The premise is the idea you believe. The joke is the specific surprise you build to prove it. "Airlines lie to you constantly" is a premise. "We're hoping to get you out as soon as possible" — said in that fake-cheerful gate-agent voice — is a joke. One is the foundation. The other is the house.

Now, the setup and the punchline you already know. The setup builds an expectation, the punchline breaks it — that's the engine. But here's where it gets more interesting, and more profitable. Once you've got a setup and a punchline that works, you are sitting on a piece of machinery that can pay you more than once. The audience is still laughing at your punchline. They haven't reset yet. And in that exact window, you can hit them again with what comedians call a tag.

A tag is an extra punchline that rides on the same setup. You did all the work building the premise and the setup — the tag gets a second laugh for free. Mint Comedy walks through a clean example. The setup is "I went to the doctor last week." The punchline is a bit of wordplay — he said you're killing it at your job, but dying at it as a hobby. That already lands. Then the tag arrives: "Which is weird, because I thought that was a hobby too." Notice what happened there. The comedian didn't build a new setup. The tag finds fresh comedic territory inside the world the punchline already created. The audience is still in that room, still laughing, and you reset the laugh by finding a second turn.

This is the part nobody mentions in the early days, and it's where the real efficiency lives. Skilled comedians get three or four tags in a row off one setup — what they call stacking tags. Think about the economics of that. You built one piece of machinery. The setup, the premise, all of it. And instead of getting one laugh and moving on, you're getting four. That's the difference between a beginner who lists jokes and a pro who builds. The beginner says one funny thing, pauses, says an unrelated funny thing, pauses. The pro stays in one spot and keeps pulling laughs out of the same ground.

Worth knowing — tags are also the easiest place to find new material on stage. You'll write a joke, perform it, and somewhere in the laugh you'll think of a line that wasn't in your notes. That line is a tag your subconscious just handed you. Keep it. Some of the best tags in any comedian's set were never written down in advance. They surfaced live, in the gap, while the room was still laughing at the punchline before it.

Now let's talk about the move that feels like magic when it works. The callback. A callback is a later joke that references an earlier joke from the same set. You said something funny ten minutes ago. The audience laughed and moved on. And then, in a completely different bit, you reference that first joke — and the room detonates. The laugh is bigger than the original was, and the strange thing is, you barely did any work for it.

Mint Comedy explains exactly why this hits so hard. When the callback lands, the listener feels recognition and surprise at the same time. Recognition, because they remember the original joke and they're proud of themselves for catching it. Surprise, because they didn't see it coming back. Those two feelings firing together is what makes a callback land harder than a fresh joke. The audience isn't just laughing at the line. They're laughing at the fact that they got it — they were paying attention, and you rewarded them for it.

But here's the constraint that trips people up, and you have to respect it. A callback can only work in live performance, or in a single continuous set, because it requires the earlier joke to still be fresh in the listener's memory. If you tell the first joke on Tuesday and the callback on Thursday, nobody remembers. The whole effect depends on the original being recent enough that the brain still has it loaded. Which means callbacks are an architecture decision. You don't stumble into them. You plant the first joke early knowing you're going to harvest it later — and we'll get deep into sequencing a full set near the end of this course. For now, just know the mechanism: plant early, pay off late, and the recognition does the work.

Here's a way to feel the difference. A tag stays inside one bit and pulls a second laugh right away. A callback leaves a bit, lets time pass, and reaches back across your whole set to connect two things the audience thought were unrelated. Tags are local. Callbacks are structural. A beginner can write a tag in an afternoon. A callback requires you to think about your set as a single shape with a beginning and an end — which is exactly the skill that separates a tight five minutes from five minutes of random jokes.

So far this has all been about words — premise, setup, punchline, the lines you write down. But a punchline doesn't only live in the words. Sometimes the funniest version of a punchline isn't something you say at all. It's something you do. And that brings us to two delivery tools that turn a written joke into a performed one: the reveal and the act-out.

The reveal is a specific kind of punchline where the twist comes from disclosing information the audience didn't have. Mint Comedy calls it the "turns out" structure. You set up a scenario, the audience assumes one thing, and then you reveal that the real situation was something else the whole time. The reason this works for longer jokes — especially storytelling — is that the punchline needs room to develop. You can't reveal something in one sentence. You have to build the false picture first, let the audience get comfortable inside it, and then pull the floor out. The longer they believed the wrong thing, the bigger the drop when you reveal the truth.

The act-out is different, and it's the one that separates people who tell jokes from people who own a stage. Instead of describing what someone said or did, you become that person for a moment. You do the voice. You do the face. You do the gesture. The Greg Dean comedy system has language for this — it talks about adopting different points of view, the Self and the Character, performing roles inside a scene. Charlie Case, one of the earliest figures credited with this kind of monologue comedy back in the vaudeville days, was doing a version of this over a century ago — talking as different people, no props, no costumes, just shifting who he was for a beat.

Here's why act-outs hit harder than description, and it's pure psychology. When you describe a scene, the audience processes words. When you act it out, they picture it. Mint Comedy nails the reason: act-outs trigger the audience to picture the scene more vividly than a described version ever could. Take a joke about your mother yelling at you. You can say "my mom got really angry" — that's a description, and it's flat. Or you can snap into your mother's posture, hit her exact tone, freeze your face into her exact look of disappointment. The second version isn't telling them she was angry. It's showing them. And the laugh is twice as big, because their imagination did the work and you just conducted it.

This is the part most beginners genuinely struggle with, and there's nothing wrong with finding it hard. Writing a clever line on paper is one skill. Physically becoming a frightened version of yourself, or a smug coworker, or a confused dog, in front of strangers — that takes stage time and a willingness to look ridiculous. The comics who can do it didn't get it from a notebook. They got it from doing it badly many times until their body learned what their pen already knew.

Now let's put all of it together, because this is the whole point. A bit isn't a list of jokes. A bit is a structure that builds. You start with one strong premise — your angle, the thing you believe. You write a setup and a punchline that proves it. Then you stack a couple of tags to wring extra laughs out of the same machinery. Maybe one of those punchlines lands better as an act-out than a spoken line, so you perform it instead of saying it. Maybe the whole thing ends on a reveal. And then, three bits later, you reach back and call one piece of it back — and the audience feels the whole set click into a single shape.

That's the difference between sounding like someone with a bag of jokes and sounding like a comedian. The bag-of-jokes person gets a laugh, resets, gets another laugh, resets. Every laugh costs them a whole new setup. The comedian builds. One premise, mined for tags, delivered with act-outs, tied together with callbacks — so the laughs compound instead of starting from zero every time. Strip it all down and it's three moves doing the heavy lifting. Find one premise strong enough to hold weight. Squeeze it for everything — every tag, every reveal, every way to perform it instead of just say it. And tie the end back to the beginning so five minutes feels like one thing, not forty things.

Which is exactly the trade-off this whole course keeps circling. The comic who looks like they're just riffing, just listing whatever pops into their head, is almost always the one who engineered the underlying structure most carefully. The next question, then, is which kind of premise fits you — because the angle a one-liner machine like Steven Wright mines is a completely different shape from the one a storyteller like John Mulaney builds on, and finding your shape is where this goes next.