Scenes are the only unit of fiction that readers actually experience in real time. Not chapters, not acts, not the story arc you outlined on index cards. The scene is where the contract between writer and reader gets enforced, sentence by sentence. A few things I've figured out about what makes them work:
Every Scene Needs a Character Who Wants Something They Can't Easily Get
Dwight Swain broke scene structure into a framework so clean it almost feels too simple. A scene, he argued in Techniques of the Selling Writer, has three moving parts: goal, conflict, disaster. A character walks in wanting something. Something gets in the way. And then things go worse than expected.
The mistake most writers make with this framework is treating it like a checklist. Goal: check. Conflict: check. Disaster: check. But Swain's real insight wasn't the three parts themselves. It was the idea that a scene without a wanting character is just description with dialogue stapled to it. The goal doesn't have to be spoken aloud. The character doesn't have to announce it. But the writer has to know what it is, and the reader has to feel it pulling the scene forward like a current under still water.
Raymond Carver's "Cathedral" is useful here. On the surface, the narrator doesn't want anything. He's sitting in his living room. His wife's blind friend is visiting. He's drinking and being vaguely unpleasant about the whole situation. If you ran Swain's framework over it like a template, you might say there's no goal. But there is. The narrator wants to not be changed. He wants to stay exactly where he is, in his comfortable smallness, in his suspicion of anything that asks him to see differently. That's a goal. It's a negative goal, a goal of resistance, but it generates just as much scene tension as a character storming a castle.
The conflict in "Cathedral" comes from the blind man's quiet refusal to be what the narrator expects. And the disaster, if you can call it that, is that the narrator ends up drawing a cathedral with his eyes closed, connected to another human being in a way he didn't want and can't explain. Swain's framework holds. It just looks different when Carver's the one building it.
The Sequel Is Where the Real Work Happens
This is the part of Swain that most people skip, and I think it's the more useful half.
After the scene (goal, conflict, disaster), Swain describes a sequel: reaction, dilemma, decision. The character absorbs what just happened. They sit with the consequences and weigh their options. And then they choose what to do next, which sends them into the next scene.
Most writers, especially early on, are drawn to the scene half. The confrontation, the door slamming, the argument escalating, the twist that changes everything. That's the fun part to write. But when you read a novel that feels rushed, that skips across the surface of its own story, what's usually missing is the sequel. The character never gets to react. The reader never gets to watch someone sit with what just happened and decide what it means.
I'm not entirely sure why the sequel is so hard to write. Part of it might be that we've trained ourselves to think of pacing as speed, when pacing is really about rhythm, the alternation between fast and slow, between happening and processing. A novel that's all scenes with no sequels reads like a montage. Things keep happening, but none of it accumulates, because nobody on the page is absorbing any of it and the reader, taking cues from the characters, doesn't absorb it either.
The sequel is where character happens. A person's character shows itself in what they do with the space between events. What do they notice first? What do they rationalize away? Which option do they refuse to even consider? The sequel, that quiet stretch of reaction and dilemma and decision, is where a character becomes a person on the page.
Robert McKee's Idea About Beats Changed How I Read Every Scene
McKee uses the word "beat" in Story to mean something specific. A beat, for McKee, is the smallest unit of change in a scene. One exchange, one moment, where the balance of power shifts from one character to another. Someone gains a small advantage. Someone loses ground. The emotional charge of the scene flips from positive to negative or negative to positive, even if only slightly.
A scene, in McKee's framework, is a series of beats. And the scene itself has an overall arc, a value charge that starts in one place and ends in another. If a scene starts with two characters in harmony and ends with them in harmony, McKee would say nothing happened. The scene didn't turn. It sat there.
Once I started reading scenes this way, counting the beats, tracking where the power moves, I couldn't stop. You pick up a novel and read a conversation between two characters and you can feel it: beat, beat, beat. She says something that puts him on the defensive. He recovers with a question that shifts the pressure back to her. She deflects with humor but loses a little ground. He presses the advantage and overplays it. She goes quiet, and the silence is its own beat, maybe the most telling one.
When a scene feels flat, when you can tell something isn't working but you can't name it, it's often because the beats have stalled. The characters are talking but the power isn't moving. Nobody is gaining or losing anything. The scene is spinning in place, generating noise without change. McKee's framework gives you a way to diagnose that. Count the beats. If the charge isn't shifting, the scene isn't earning its place.