The self-editing checklist: structure, then scenes, then lines
The most expensive mistake in self-editing isn't missing a flaw — it's polishing prose in a chapter you later delete. Order is everything: fix the skeleton before the muscles, the muscles before the skin. Here's the three-pass system, what belongs in each pass, and — just as important — the signal that you've hit the limit of what you can see alone.
First: let it cool, then read it cold
Before any pass, put the manuscript in a drawer for two to four weeks — long enough that you start forgetting your own sentences. Then read the whole book fast, in as few sittings as possible, the way a reader will, taking notes but changing nothing. You're not editing yet; you're building the diagnosis. Mark where you got bored, where you got confused, where you skimmed your own book. Those flinches are the truest data you'll ever get from yourself.
Resist the itch to fix as you read. A fix made before you've seen the whole picture is a fix made against the wrong picture — the sagging middle might need a subplot cut, not a livelier chapter 14, and you can't know that on page 200 of 380.
Pass one — structure (the wrecking-ball pass)
The structural pass asks the big, brutal questions, and answers them with moves measured in chapters: cut, merge, reorder, add. Does the story start in the right place, or is chapter 1 really throat-clearing before the chapter 3 inciting incident? Does the protagonist drive events, or get dragged by them? Does the middle escalate or circle? Does every subplot pay off or earn its cut? Does the ending answer the question the opening asked?
Do all of this before touching a sentence, because this pass deletes sentences by the thousand. A useful discipline: write a one-line summary of each chapter's job in the book. Any chapter whose job you can't state — or whose job duplicates another's — is a merge or cut candidate. This is also the pass where you hunt plot holes systematically: motivation, logistics, knowledge, timeline. It's the hardest pass to do alone, since it requires seeing the whole book at once, which is exactly what author-memory is worst at.
Pass two — scenes (the pacing pass)
With the skeleton set, work scene by scene. Every scene must change something — a character's situation, knowledge, or relationships; if the state of the story is identical after the scene, the scene is a candidate for cutting no matter how well-written it is. Check entrances and exits: enter as late as possible, leave on the turn, not the wind-down. Check that conflict — outer or inner — is present on nearly every page, and that scene-level tension connects to the book-level stakes.
This is also where POV and continuity live. One scene, one head. Verify the reader always knows where they are and when. Confirm objects, injuries, weather, and who-knows-what carry correctly from scene to scene — the small continuity slips that survived the structural pass get caught here, while you're already holding each scene up to the light.
Pass three — lines (the polish pass)
Only now do you sweat sentences, because only now are you sure these sentences are staying. The line pass is a bundle of hunts: filter words and distancing constructions; adverbs propping up weak verbs; your personal crutch words (everyone has three — find yours with a frequency search); dialogue tags fancier than 'said'; echoes, where a distinctive word repeats within a page; and sentences that all share one rhythm. Read dialogue aloud — the ear catches stilted lines the eye forgives.
Work in chunks with breaks; line-editing accuracy collapses when you push tired. And timebox it. Line polish is the pass that can absorb infinite hours for asymptotically small gains, and it's where perfectionism goes to hide from the scarier question of whether the book works.
When to stop: the hand-off signal
You're done self-editing when your changes stop being improvements and start being alternatives — when you're swapping a good sentence for a different good sentence, or reversing yesterday's edit. That's not a quality problem; it's an information problem. You've extracted everything your own eyes can see, and the remaining flaws are precisely the ones invisible to you: the twist you think is subtle but is actually opaque, the character you find charming but readers find grating.
That's the moment for beta readers — not when the book is perfect, but when it's stable and you've gone blind to it. Send it clean through the line pass (readers tripping on typos can't see the story), ask three to five specific questions instead of 'thoughts?', and make it easy to give feedback where it's actually usable. This is the loop Scribegrove's sharing links are built for: your reader opens a watermarked link in their browser — no account needed — and their notes come back anchored to the exact chapter and passage, instead of scattered across five email threads in five formats. Collect the feedback, look for patterns (one reader's opinion is an opinion; three readers stumbling in the same chapter is data), and run one more targeted pass. Then it's an editor or a publish button — but that's the next guide.
Frequently asked
How many editing passes does a novel need?
Three focused passes — structure, scene, line — beat ten unfocused rereads, because each pass looks for one class of problem. Some authors split further (a dedicated POV pass, a dialogue pass), which is fine as long as the order holds: never polish lines before the structure is locked.
Should I hire a professional editor or use beta readers first?
Beta readers first, almost always. They're free-to-cheap, they catch the reader-experience problems (confusion, boredom, disbelief) that would otherwise consume expensive editorial hours, and their feedback tells you what kind of professional edit you actually need — developmental if structure wobbles, copyedit if it's solid.
This guide is general information for authors, not legal advice. Platform and store policies change — verify the current terms wherever you publish.