A sane revision process

The three layers, in order

Revise in this order. Do not mix layers. Do not polish sentences while you're deciding whether a chapter belongs in the book.

Layer 1: Structure

Is the book the book? This pass happens after a complete first draft, ideally after at least two weeks of not looking at the manuscript. You read the book — not as a writer, as a reader — and ask:

• Who is the protagonist, and what does she want? Does the book track that want?
• What is the central conflict, and when does it take hold? Is it too late?
• What is the midpoint shift, and does it actually raise the stakes?
• What is the climax, and is it earned by the choices the protagonist has made?
• Does the ending address the opening? Does the book feel structurally complete?
• Which scenes are load-bearing? Which exist mostly because they got written?

Structural revision often requires cutting or rewriting tens of thousands of words. Do it now, not later. Every hour you spend polishing a scene that's about to be cut is an hour wasted.

The output of this pass: a revised outline of what the book actually is, a list of scenes to cut, scenes to add, and the beat where each new scene will land.

Layer 2: Scene

Now that the structure works, each scene has to. This is a scene-by-scene pass. For each scene:

• Whose POV is this, and is it the right POV for what this scene needs to accomplish?
• What does each character want in this scene? Are those wants in conflict with each other?
• Where is the scene's turn — the moment the situation changes?
• Does the scene end in a different place than it started, either for the plot or for the characters?
• Can the scene start later? End earlier? (Most scenes can.)
• Is there a beat here that doesn't belong to this scene — a reveal that is landing too early, a piece of information that should be in a different scene?

This pass often involves moving scenes around, compressing two scenes into one, or splitting one scene into two. Rewrite rather than patch. A scene that isn't working usually needs to be rewritten from a blank page with the scene's purpose in mind, not tinkered with.

Layer 3: Line

Only when structure and scenes are solid: line editing. This is the pass most writers jump to first. It is the last pass for a reason — it is the easiest one to redo if the book changes.

• Sentence-level rhythm. Variety of sentence length. Clusters of short sentences, clusters of long. Read aloud.
• Word-level precision. Is every abstract noun necessary? Is every verb carrying its weight?
• Dialogue tags and adverbs. Cut what can be cut. (See dialogue guide.)
• Showing versus telling at high-leverage moments. (See show, don't tell.)
• Clarity of sensory detail at scene openings.
• Unintentional repetition — of words, of phrases, of sentence shapes.

A line pass moves slowly. A thousand words an hour is normal. Don't try to line-edit a whole novel in a weekend; the attention required can't be sustained.

Beta readers — when and how

Beta readers are other readers who read your manuscript and report on their experience. Not editors. Not line-editors. Readers. They belong somewhere between layer 1 and layer 2 — after you've done the first structural pass yourself, before you commit to the scene-level work.

Give each beta reader specific questions, not a general "what did you think?" Useful questions:

• At what point did you want to put the book down? (Be ready to hear an answer.)
• Which character did you care about most, and when?
• Which sections felt slow? Mark them.
• What did you think was going to happen that didn't? What did happen that surprised you?
• If you had to cut one chapter, which?

Two or three beta readers with specific feedback is worth more than ten with vague praise. Weigh feedback by whether multiple readers flag the same thing. A single reader's opinion, however strongly expressed, is not a mandate. A flag from three readers about the same chapter is.

Knowing when to stop

Revision doesn't end by itself. Writers stop revising either because an external deadline forces them to, or because they've made a deliberate decision. A working heuristic:

• You are making changes that undo previous changes. If revision N is putting back a scene you cut in revision N−1, you are no longer improving the book, you are rotating around it.
• Beta readers' feedback stops agreeing with each other. When three readers each want a different version of the book, the book is done. Pick the version that matches your own vision and stop.
• You've been on the line pass for more than six weeks. Line editing has diminishing returns. If you've rewritten chapter three seven times, chapter three is done.
• Submitting is scary, which is why you keep revising. This is the hardest signal to catch in yourself. If the honest answer to "why am I still tinkering" is "because if I send it and it's rejected, that will feel bad," it is time to send it.

After querying starts

Once you are actively querying, stop revising the manuscript based on query rejections — unless you see a very clear pattern of the same feedback from multiple agents. Tinkering with the manuscript every time a form rejection arrives produces a book that chases rejections rather than one that has been decided on. If you feel the urge to revise, start the next book instead.