The Two-Question Scene Test
Most dead scenes don't fail because the prose is bad. They fail because nothing is at stake and nothing moves. A character walks into a room, describes the room, has a pleasant conversation, and leaves exactly the person they were when they walked in. The reader can't tell you what the scene was for, and neither can you — because it wasn't for anything.
THE SHORT ANSWER
Before you write any scene, answer two questions: who wants what in this scene, and what changes by the end of it? If you can't answer both, you don't have a scene yet — you have a location where characters happen to be standing around. Fix the want or the change, and the scene usually fixes itself.
The Problem
The usual advice for this is "add conflict," which is the kind of advice that sounds helpful and helps no one. Conflict with what? About what? A shouting match is conflict, and it can still be a dead scene if nobody wants anything specific and nothing changes when it's over.
So here is a sharper test. Two questions, asked before a single line is written:
- Who wants what in this scene? Not in the story — in this scene, right now, in this room. The want should be concrete enough to be denied.
- What changes by the end? A decision made, a secret learned, a relationship shifted, a door closed. Something must be different on the last line than it was on the first.
A scene that passes both questions has an engine. A scene that fails one of them has a diagnosis — and a repair.
The Worked Example: A Scene That Fails Both
Here is a scene I made up to show you the failure mode. Mara, a junior cartographer, visits her old mentor, Tomas, in his workshop. Read it and notice how pleasant — and how inert — it is:
Mara arrived at Tomas's workshop just as the afternoon light came through the high windows. The workshop smelled of ink and old paper, and maps covered every wall — coastlines, mountain ranges, harbors she had once helped him draw. Tomas made tea. They talked about the old days, about the survey expedition that had made both their names, and about how the coastline had changed since. He showed her his newest chart, a beautiful thing, all careful inkwork. She admired it. When the tea was gone, she thanked him and left.
Lovely. Also: nothing happened. Ask the two questions:
- Who wants what? Mara doesn't want anything in this scene. She's visiting. Tomas doesn't want anything either — he's showing off a map and pouring tea. Nobody is trying to get anything from anyone.
- What changes? Nothing. Mara leaves with the same knowledge, the same relationships, and the same plans she arrived with. If you cut the scene, the story would not notice.
This is the trap: the scene feels like writing because it contains description, backstory, and a character the writer likes. But description is not action, backstory is not stakes, and a likable character standing still is still standing still.
The Repair
Keep the setting. Keep the tea. Give Mara a want and give the scene a change.
Suppose Mara has come because the new harbor chart — the one Tomas is so proud of — contains an error she spotted, one that will send ships onto rocks if it goes to the printers tomorrow. But Tomas is the mentor who taught her everything, and telling him means accusing the great man of a mistake in front of his life's work. Now the scene has a want: Mara wants Tomas to pull the chart before it prints, without destroying his pride or her standing with him. And Tomas has a want of his own: he wants his protégée's visit to be what it looks like — a fond reunion, not an inspection.
Same room, same tea, completely different scene:
Mara arrived at Tomas's workshop just as the afternoon light came through the high windows. The maps on the wall were the ones she had once helped him draw — coastlines, mountain ranges, the harbor. Tomas made tea, and for a while they talked about the old days, the way people do when they are both avoiding the real subject. Then Mara set her cup down and said, "Tomas, the new harbor chart. The eastern approach — I measured it twice." Tomas went very still. "You came all this way to correct my chart?" She could have said yes. Instead she said, "I came all this way because no one else would dare to." He looked at the chart, then at her, and for the first time since she arrived, he really saw her — not his student, but his equal. "Show me," he said.
Now ask the two questions again:
- Who wants what? Mara wants the chart pulled without humiliating her mentor. Tomas wants the visit to stay a reunion. Those wants collide, which is what makes the conversation worth reading.
- What changes? The relationship changes. Tomas stops seeing Mara as his student and starts seeing her as his equal — and the chart's fate is now an open question instead of a settled fact. Cut this scene and the story does notice.
Notice what the repair did not require: no new characters, no explosion, no plot twist. It required one want, aimed at one obstacle, with one change at the end. That is the whole trick.
TRY THIS TODAY
Take the scene you are most unsure about in your current draft — the one you keep skipping past when you reread — and run the test on it in the margin:
- Write, in one sentence: who wants what in this scene?
- Write, in one sentence: what is different on the last line than on the first?
If you can't write either sentence, don't revise the prose yet. First, give somebody a want: something they could plausibly ask for, be denied, or be afraid to ask for. Then decide what changes: a decision, a revelation, a shift in how two characters see each other. Rewrite the scene around those two sentences, and only then polish the sentences themselves.
One warning, from hard experience: the want has to belong to the scene, not to the novel. "Mara wants to save the kingdom" is a story-level want, and it will not power a tea scene. "Mara wants Tomas to hear her out before the printers run tomorrow" will. Keep the want small enough to fit in the room you're writing.
This is the fourth pillar doing its quiet work: dramatic cause and effect. A scene is not a string of things that happen one after another — "and then, and then, and then." It is a chain where one thing causes the next: a want meets an obstacle, and the collision changes something. The two-question test is just a pocket-sized way to check whether your chain has any links in it.
If a scene passes both questions, write it with confidence. If it fails one, you now know exactly what to fix — and that, frankly, is the whole job.
The takeaway: If a scene passes both questions, write it with confidence. — See the Method.