Why Your First Draft Is Supposed to Be Terrible
You finished the draft. You read it back. It’s awful. The pacing is wrong, the dialogue is stiff, the middle sags, three subplots go nowhere, and your protagonist does something in chapter twelve that contradicts everything established in chapter four.
I love revision, and a writer who treats a bad first draft as proof they can’t write is misjudging the process. A lousy draft is the normal state of every book at that stage, structurally and mechanically. The writers who quit over it are quitting over the one part of the process that goes wrong for everybody, and they’re usually a few months of honest work away from a book.
First drafts are discovery documents. You were learning your story while you wrote it. Of course the early chapters don’t match the late chapters. You didn’t know where you were going when you started.
I love revision, and a writer who treats a bad first draft as proof they can’t write is misjudging the process.
I’ve written 113+ books. Every single one had a terrible first draft. The difference between published authors and abandoned manuscripts is willingness to revise systematically instead of tinkering endlessly or giving up entirely.
The draft exists so you can revise it. Revision is where the book becomes itself.
Watch: Sanderson 2014.7 - Revisions and Prose, zmunk
What should you fix first when revising a novel?
Most writers start revision by fixing sentences, polishing prose and fiddling with word choice, and I think it’s the most expensive mistake in the whole process. It’s repainting the walls of a house with a cracked foundation. You can lose months to it, and the worst part is how productive it feels while you’re doing it.
Revision works in layers, from large to small. (See also James Scott Bell’s Revision and Self-Editing) Fix structural problems first. Then character consistency. Then scene-level pacing. Then dialogue. Then prose. Then line edits. If you polish sentences in chapter three and later discover that chapter three needs to be cut entirely, you’ve wasted time on work that doesn’t survive. The hierarchy looks like this: Structural revision: Does the plot work? Do the acts function? Is the inciting incident in the right place? Does the midpoint shift the story meaningfully? Does the climax justify the ending?
Character revision: Are characters consistent? Do their decisions follow from their psychology? Does the wound-adaptation-pattern hold across every scene? Does the arc land?
Scene revision: Does every scene turn? Is the emotional value different at the end than at the beginning? Can any scene be cut without affecting downstream scenes? If yes, cut it.
Dialogue revision: Does each character have a distinct voice? Is there subtext? Can you remove the tags and still identify who’s speaking?
Prose revision: Sentence-level polish. Rhythm. Word choice. Eliminating filler words, passive constructions, and the dead language that crept in during drafting.
The Revision Pass Checklist
Work the passes in order, because fixing a lower level first wastes effort if the scene is later cut.
Each revision pass has specific diagnostic questions. Work through them in order and don’t skip ahead. Fixing a lower-level problem before the higher-level pass is wasted effort if the scene later gets cut or restructured, and nobody enjoys finding out they spent a week polishing a chapter that was headed for the trash.
Pass 1. Structure (allow 1-2 weeks): Read the manuscript in one sitting, taking notes only. Mark where your attention drops. Then ask: Does the inciting incident happen within the first 15% of the book? Does the midpoint change the protagonist’s approach to the problem? Does every subplot connect to the main arc? Can you state the climax in one sentence? Does the ending deliver on the promise of the opening? Cut or relocate any chapter that doesn’t advance the central conflict.
Pass 2. Character (allow 1 week): Track each named character through the manuscript. Ask: Does this character want something in every scene they appear in? Does their behavior follow from their wound and adaptation? Do they change between their first and last appearance? Can any two characters be combined without losing plot function? Does the protagonist make the climax happen through their own choice, without luck or rescue?
Pass 3. Scene (allow 1 week): Examine each scene individually. Ask: What is the emotional value at the start of this scene? Is it different at the end? If this scene were deleted, would any downstream scene need to change? Does the scene enter late and exit early? Is the point-of-view consistent throughout?
Pass 4. Dialogue (allow 3-5 days): Read all dialogue aloud. Ask: Can you identify the speaker without tags? Is there subtext, a gap between what the character says and what they mean? Are characters exchanging information for the reader’s benefit or talking the way real people talk? Does every conversation contain at least a hint of conflict or tension?
Pass 5. Prose (allow 3-5 days): Line-level polish. Ask: Can any word be cut without changing the meaning? Are you using active voice unless you deliberately choose passive? Are there filler phrases (“began to,” “seemed to,” “started to”) that you can eliminate? Does the sentence rhythm vary? Does the prose carry any AI-generated language patterns that you need to clean?
The Revisions Handbook walks through each level of this hierarchy with checklists, diagnostic questions, and AI prompts for identifying problems at each layer before you start fixing them.
How to Evaluate Your Own Work
The hardest part of revision is seeing what’s broken. You’re too close to the manuscript. You know what you meant, so you read meaning into passages that don’t deliver it. You remember the emotional context of a scene, so you feel the emotion even when the prose doesn’t create it. Distance helps. Put the manuscript away for at least two weeks before starting revision. A month is better. You need enough time to forget the specifics so you can read what’s on the page instead of what’s in your head.
The fix is to make your own words look unfamiliar. A writer in a creative writing group whose draft had blurred together after too many reads wanted to rewrite the whole book by hand to see it fresh. That works for some people, but it’s weeks of copying, and you still have to type it all back in and proofread it again.
Other writers in the thread had cheaper tricks. Change the font and the text color before you edit, so the lines break in new places and your eye stops skimming. Convert the file to an ebook and read it on a different device. Let text-to-speech read it to you while you follow along. One writer marks every scene in alternating highlight colors and reads them out of order, so the flow of the story can’t carry her past the small problems.
My own habit works at the chapter level. Before I start a new chapter, I reread the one before it, and sometimes I go back two. By the time I get to chapter two, it’s probably been a couple of weeks since I read chapter one, and I want it fresh. I find a lot of stuff rereading like that, and most of what I fix is flow.
Read the entire manuscript in one or two sittings without stopping to fix anything. Mark problems with a simple notation (“slow,” “confusing,” “who cares?”) and keep reading. The goal is to experience the manuscript the way a reader would: continuously, without the author’s backstage knowledge.
Your gut reactions during this read-through are the most valuable revision tool you have. Where you got bored is where readers will get bored. Where you got confused is where readers will get confused. Where you felt nothing is where readers will stop reading. Trust the feeling even when you can’t explain it yet.
Cutting Scenes You Love (And Why It Makes the Book Better)
The most common revision failure is refusing to cut material that doesn’t serve the story. You wrote a beautiful scene. The prose sings. The dialogue crackles. It took you three days to get it right. And it doesn’t advance the plot, develop characters, or increase tension.
Cut it. I’d rather lose a scene I love than keep one that makes a reader put the book down in the middle, because that reader doesn’t come back for your next book either. The scene can be beautiful and still be wrong for this book, and a charming character can still be one character too many.
Every element in the final manuscript has to justify its place by doing something nothing else in the book does. The test is simple: remove the element and read the surrounding pages. If the story flows better without it, the element was interrupting instead of contributing. If nothing downstream breaks, the element wasn’t essential. Save your cuts in a separate file. You’re not destroying them. You’re moving them somewhere they can’t damage the pace, structure, and focus of the book you’re writing. Some of those cuts might become the seed of your next project.
Recognizing and Fixing Awful Writing Patterns
Every writer has default bad habits, patterns they fall into during drafting that need to be identified and corrected during revision. The faster you learn your personal patterns, the faster your revision process becomes.
Common awful writing patterns include hedging language (she seemed to maybe feel somewhat upset), filter words that create distance (he noticed that the room was dark vs. the room was dark), echoed words and phrases repeated within the same page or paragraph, dialogue that summarizes emotion instead of showing it, and scenes that start too early or end too late.
AI-generated text has its own awful patterns: corporate transitional phrases, hedging qualifiers, balanced-to-the-point-of-saying-nothing conclusions, and the word “delve.” If you’ve used AI during drafting, cleaning up those AI fingerprints is a distinct revision pass on its own. The Awful Writing Handbook catalogs the most common bad writing patterns across prose, dialogue, structure, and AI-assisted drafting. Each pattern includes examples, diagnostic tests, and AI prompts designed to identify instances in your manuscript so you can fix them systematically instead of hoping you catch them on a read-through.
When should you use beta readers?
Beta readers are essential, after you’ve finished your own structural revision. (See why beta readers can’t diagnose a first novel)
Sending them a first draft burns goodwill you’ll need later, and I think it’s unfair to them. They’ll flag structural problems you already know about and react to character inconsistencies you haven’t fixed yet. Their feedback will be accurate and useless, because it’s about a version of the book you were going to tear apart anyway.
Send beta readers a manuscript that represents your best effort. You’ve done the structural pass. You’ve checked character consistency. You’ve fixed pacing. You’ve polished the prose. Now you need eyes that aren’t yours to tell you what still isn’t working. Good beta reader questions are specific. “Did the midpoint surprise you or did you see it coming?” is useful. “What did you think?” isn’t. “At what point did you first consider putting the book down?” gives you practical data. “Did you like it?” gives you nothing.
Not all beta reader feedback is equal. If one reader found chapter eight confusing, it might be their reading. If three readers found chapter eight confusing, chapter eight has a problem. Look for patterns in the feedback.
The Revision Timeline: How Long This Takes
First-time authors underestimate revision by a factor of three to five. If your first draft took four months, expect revision to take two to four months at minimum. More if the structural problems are significant.
That gap is where a lot of books die. A writer who budgeted two weeks hits month two, decides something’s wrong with them, and abandons a manuscript that was on a perfectly normal schedule. It’s a waste, because nothing was wrong except the expectation somebody handed them.
A realistic revision timeline looks like this: two to four weeks of rest between finishing the draft and starting revision.
One full read-through with notes, taking one to two weeks. Structural revision, two to four weeks. Character and scene revision, two to three weeks. Dialogue and prose polish, two to three weeks. Beta reader period, four to six weeks. Post-beta revision, two to three weeks.
| Stage | Time to allow | What it involves |
|---|---|---|
| Rest the draft | Two to four weeks | Distance so you read what is on the page, not what is in your head |
| Full read-through | One to two weeks | Read without stopping; mark “slow,” “confusing,” “who cares?” |
| Structural revision | Two to four weeks | Cut or relocate any chapter that does not advance the central conflict |
| Character and scene revision | Two to three weeks | Characters want something in every scene; every scene turns |
| Dialogue and prose polish | Two to three weeks | Read dialogue aloud; cut filler phrases and dead language |
| Beta reader period | Four to six weeks | Send your best effort and ask specific questions |
| Post-beta revision | Two to three weeks | Act on patterns in the feedback, not one reader’s reaction |
This isn’t a production line. Stages overlap. You’ll discover structural problems during prose polish. You’ll fix dialogue during character revision. The hierarchy guides your focus and doesn’t set a rigid schedule. But knowing the real timeline prevents the frustration of expecting to “quickly clean up” a draft and finding yourself still revising months later.
My take
Structure goes first and prose goes last, and doing it backwards wastes months. Expect revision to take a long time. Writers who budget two weeks hit month two, decide something’s wrong with them and walk away from a manuscript on a perfectly normal schedule. The only fault was the expectation. Rest the draft, read it through, then work the passes in order. Give it the time and you’ll finish.
When should you stop revising a novel?
Revision has a point of diminishing returns. After a certain number of passes, you’re not improving the manuscript. You’re rearranging furniture. Changing a word back and forth between drafts. Cutting a sentence and then putting it back. Adding a comma and then removing it.
The manuscript is ready when you can read it without stopping to fix things. Not without noticing imperfections (every author sees those forever), but without feeling that the imperfections are structural, character-based, or pacing-related. If your objections are all at the sentence level and even those are minor, the book is done.
I think perfectionism dressed up as revision is procrastination with better excuses, and it costs writers more than any bad review ever will. The book doesn’t need to be perfect. It needs to be finished, polished and honest, and a writer who keeps it in a drawer for a sixth pass is keeping it from readers who’d have been glad to have it a year ago.
AI can make the loop worse. A writer in an AI authors’ group co-wrote a chapter with a chatbot, read it, marked changes and rewrote it. Then new problems turned up, and the cycle started over. Every pass raised more what-ifs, and the book never got closer to done. The tool will happily find something to change forever, because finding something is what you asked it to do.
The writers who’d broken out of that loop gave the same advice. Finish the whole draft before you polish any of it. Take a break before you edit. Fix one kind of problem per pass, so a pass for dialogue isn’t also a pass for pacing and description. And when you ask AI for help, ask for a list of problems instead of a rewrite, then make the changes yourself.
I use AI two ways at that stage. I give it the whole novel and ask about structure: which characters I introduced and then forgot, and which stretches run slow. Then I work chapter by chapter for the smaller notes. The whole book shows you the structure, and a single chapter gives you detail you can act on. Neither one will tell you the book is done. You decide that.
A published imperfect book reaches readers. A perpetually revised manuscript reaches nobody. The AI-Enhanced Writer’s Library covers revision as an integrated craft skill, connecting structural revision to character work, pacing, dialogue, and prose polish so each revision pass addresses multiple layers at once.
