Most scenes don’t fail because the writer can’t write.

They fail because the writer sat down, started the scene, and never decided what the scene was supposed to do. The prose is fine. The dialogue is fine. There’s nothing in the scene you could point at and call broken. It just doesn’t move the book forward, and you can’t say why.

The reason is almost always that the scene didn’t answer all five of the questions a working scene has to answer. Get one wrong and the scene wobbles. Get two wrong and the scene cuts in revision. The five questions are old craft, but they’re rarely written down in one place.

1. Whose scene is it?

The point-of-view question. Not “whose name is in the chapter heading” — whose interiority is the engine of this scene? Whose stakes are we tracking? Whose decisions matter?

In single-POV books this is settled before you sit down. In multi-POV books it’s the first decision you make, and it’s a real decision. The same event told from two different POVs is two different scenes — different information, different stakes, different turn. Decide before you write. Re-decide if the scene wants to drift to a different character.

2. What does the POV character want, in this scene?

Not what they want in the book. What they want right now, on this page. Specific, immediate, achievable-or-not within the scene.

A protagonist’s book-level want — find the killer, save the kingdom, win the trial — is too big to drive a single scene. Each scene needs a smaller want nested inside the big one. Get the address out of the witness. Convince the queen to grant an audience. Survive the next ten minutes without breaking. Concrete enough that you can check, at the end of the scene, whether they got it.

If you can’t name the scene-level want, the scene is wandering.

3. What’s stopping them?

The conflict question. Could be a person, a circumstance, an internal block, a clock — but something has to push back, or the scene is just admin. Things happening in sequence is not a scene. Things happening against resistance is a scene.

Most flat scenes flatten here. The protagonist wants something, they get it, the scene ends. No friction means no movement. If your protagonist is cruising through, you either need to up the resistance or cut the scene and skip to the one where resistance shows up.

4. What changes by the end?

The turn. The state of the protagonist, the situation, or the reader’s understanding has to be different at the end of the scene than at the start. This is non-negotiable. Scenes that don’t turn are scenes that revision will eventually delete.

The change can be small. A new piece of information. A relationship recalibrated by one degree. A door closed that was open at the start. But something must move, and you should be able to write the change in one sentence: Sarah went into this scene believing X; she leaves believing Y.

If the only change is “they had a conversation,” the scene hasn’t earned its place.

5. What does the reader know now that they didn’t know before?

The information transaction. Every scene is a contract with the reader: in exchange for the time they spent on this page, they get something they couldn’t have gotten anywhere else. New facts, new emotional context, new texture, new tension. Something.

Scenes that re-cover ground the reader already covered are scenes the reader skips. Scenes that withhold information the reader has earned are scenes the reader resents. The information transaction is what keeps the contract honest.

This question also doubles as a backstop on the other four. If you can’t answer “what did the reader learn here,” there’s a good chance the scene was wandering, the conflict was thin, or the turn never happened.

What this looks like in practice

Open any scene in your current manuscript. Five sticky notes. Answer the five questions on them, in order, in one sentence each.

If you can’t answer one of them, the scene has a structural hole. That hole is what’s making the scene feel weak even when the prose is clean.

This is also the diagnostic to run on a chapter that you know isn’t working but can’t fix. The fix is almost never “tighten the prose.” The fix is almost always “identify which of the five questions wasn’t answered, and answer it.”

Why we tag scenes against this in Author

Storyverse Author’s scene tracking is built around exactly this kind of question-by-question audit. POV character, scene goal, scene conflict, outcome, and information delivered are first-class fields on every scene — not because we invented them, but because they’re the questions that matter and most tools don’t bother to surface them.

The point isn’t to fill in fields. The point is that filling them in forces the decision before you write the prose, instead of after revision tells you the prose was hollow.

What to do this week

Pick three scenes from your current draft. Don’t pick obvious problem scenes — pick three that you think are fine. Run them through the five questions.

The scenes you thought were fine will fail one or two of the questions more often than you’d expect. That’s the lesson. The questions aren’t a punishment. They’re a backstop. Use them before you write, and you’ll write fewer scenes that revision has to throw out.

Get the five questions right and the rest of the scene becomes a craft problem — pacing, prose, dialogue. Solvable problems. Get them wrong and you’re polishing a scene that doesn’t know what it’s for.