The arc of an improv scene — not a rigid template but a recognizable shape that emerges when the principles are working.
Three phases:
1. Establish (base reality). Who are these people? Where are they? What are they doing? What's the relationship? One to three lines plus a shared physical activity. Jimmy Carrane's advice: start "in the middle" — skip formalities and begin with a statement that addresses what's already happening in the relationship. Don't small-talk your way in.
2. Discover and play (game). Something unusual surfaces. Both players recognize it. The scene shifts from establishing to exploring — heightening the pattern, finding new contexts, deepening the emotional stakes underneath. This is where the scene lives. The middle is not a single event but a cycle: heighten → explore the new reality → heighten again.
3. Resolve (edit/button). The scene ends. In improv, endings are usually chosen, not scripted:
- The button — a late endowment or reversal that recontextualizes everything. Gets a big response as the audience reprocesses.
- Edit on the laugh — a sweep edit at the peak of a heighten. Leave them wanting more.
- The rule of three — if the game has hit three beats, it's likely done. Variations exist, but three is the natural rhythm.
- The blowout — the pattern escalates to a point of absurdity that transcends itself.
When is a scene "done"? Replace "someone should edit this" with "I should edit this." The best time to edit is almost always before you think it's time. A scene that runs too long decays; a scene that's cut slightly early leaves energy for the next.
Two structural traditions:
- UCB/game-first: Base reality → first unusual thing → if-this-then-what → heighten-and-explore → button/edit. Structure comes from game mechanics.
- Johnstone/narrative-first: Platform → tilt (interruption of routine) → reincorporation. Structure comes from story. Johnstone: "Set a firm platform, expand it, and reincorporate like mad. You don't know where you're going, just where you've been."
Both produce scenes with the same shape (establish → explore → resolve). They differ in what drives the middle: game pattern vs. narrative consequence.
Scenes need change. A scene where nothing shifts is a premise, not a scene. The relationship should be different at the end than at the beginning — emotionally, in terms of status, in terms of what the characters know about each other. Without movement, scenes decay.