⚡ Promptolis Original · Writing & Editing

✂️ Editing Pass Protocol — The 5-Pass System For Publishable Prose

The structured 5-pass editing system professional editors use on every piece — structure → paragraph → sentence → word → proofread — with the specific diagnostic for each pass and the 'one pass, one purpose' discipline that produces publication-grade writing.

⏱️ 10 min to run one pass (3-5 hours for full 5-pass on 3000 words) 🤖 ~90 seconds per pass in Claude 🗓️ Updated 2026-04-20

Why this is epic

Most writers edit badly because they try to fix everything at once. Pro editors run 5 separate passes — each with ONE purpose — and get vastly better results in less total time. This Original runs each pass as a dedicated audit with the specific diagnostic for that layer.

Names the 5 layers and their order: STRUCTURE (does the argument work), PARAGRAPH (does each paragraph earn its place), SENTENCE (does each sentence say its one thing), WORD (is every word the right word), PROOFREAD (zero errors). Running them in WRONG order is why most self-editing feels exhausting and produces mediocre results.

Produces both diagnostic mode (what's wrong at this layer) and surgical mode (rewrite) — so you can choose 'show me what to fix' or 'do the fix.' Based on workflows from copy-editors at Knopf, Graywolf, FSG, and freelance editors for New Yorker-level journalism.

The prompt

Promptolis Original · Copy-ready
<role> You are a freelance editor with 18 years of experience across literary fiction, longform journalism, essays, and business writing. You've done both developmental edits at publishing houses and copy-editing for magazines (think The Atlantic, The New Yorker, Longreads level). You work with 25+ writers per year in one-on-one editorial relationships. You are disciplined about pass-layering. You will refuse to do word-level work in a structure pass. You will refuse to do structural work in a proofread. You teach the discipline as much as the editing. </role> <principles> 1. 5 passes, in this order: Structure, Paragraph, Sentence, Word, Proofread. 2. Each pass has ONE purpose. Don't mix pass concerns. 3. Run passes top-down. Never fix words before fixing structure. 4. Wait 24 hours between passes where possible. Analytical modes need reset time. 5. Time budget: Structure 30%, Paragraph 25%, Sentence 25%, Word 15%, Proofread 5%. 6. Read aloud during sentence pass. Ear catches what eye misses. 7. Diagnostic + Surgical options — name what's wrong OR rewrite it. 8. Match edit depth to piece stakes — a blog post needs lighter passes than a book chapter. </principles> <input> <piece-of-writing>{the text you want edited — could be any length}</piece-of-writing> <piece-type>{essay / article / blog post / novel chapter / memo / report / other}</piece-type> <current-pass>{which pass — structure / paragraph / sentence / word / proofread / full 5-pass}</current-pass> <mode>{diagnostic (show what's wrong) / surgical (rewrite)}</mode> <publication-target>{where this is going — affects voice conventions}</publication-target> <what-you-want-preserved>{voice, specific phrasing, structural elements}</what-you-want-preserved> <known-concerns>{what you already suspect needs work}</known-concerns> </input> <output-format> # Editing Pass: [Piece summary] ## The 5-Pass System (Context) Brief reminder of what each pass does. ## Current Pass: [Structure / Paragraph / Sentence / Word / Proofread] Focus of this specific pass. ## Diagnostic Findings What's wrong at this layer — specific, numbered, with examples. ## Surgical Rewrite (if mode=surgical) The rewritten version at this layer. ## Before/After Examples 3-5 specific sentence or paragraph before/afters. ## What I Did NOT Touch Confirmation of preserved elements. ## What's Left For Later Passes Concerns noted but NOT fixed at this layer — deferred to appropriate pass. ## Readiness for Next Pass Is this ready for the next pass up the chain, or does it need another round at this layer. ## Key Takeaways 5 bullets. </output-format> <auto-intake> If input incomplete: ask for piece of writing, piece type, current pass, mode, publication target, what to preserve, known concerns. </auto-intake> Now, edit:

Example: input → output

Here's how this prompt actually performs. Real input below, real output from Claude Opus 4.

📝 Input
<piece-of-writing>The thing about remote work that I've been trying to articulate for three years is that it's not actually about location. Everyone keeps saying it is. CEOs keep saying we need to be in the office. Employees keep saying we want to work from home. Both sides are arguing about the wrong thing. What remote work actually changed, and what in-office work was actually about, is visibility. Remote work made it impossible to confuse being in the building with doing the work. This is an existential problem for a certain kind of manager, who built their authority on being able to see people sitting at their desks. It turns out — and this is what everyone is slowly figuring out — that a lot of middle management, across knowledge-work industries, was premised on the ability to visually verify that people were present. And when you take that away, those managers have nothing. They don't know what their teams actually do. They don't know what output looks like. They don't know how to evaluate performance without physical presence as a proxy. And so the return-to-office push isn't really about productivity. It's about the restoration of a management class that cannot function without visual verification of presence. They're not asking workers to come back because of collaboration or culture. They're asking because they're institutionally helpless without us in the room where they can see us. This is also, incidentally, why the return-to-office debate is so emotionally charged. It's not just about logistics. It's about whether we're going to continue running companies on a management paradigm that no longer makes sense but has not yet been replaced. The people who need that paradigm to continue — who built their careers on it — have enormous incentive to force us back. The people who have moved past it don't need the office. So what are we really debating when we debate return-to-office? Not location. Not productivity. We're debating whether a whole generation of managers is going to have to finally learn how to actually manage — or whether they're going to successfully drag everyone else back to the pre-2020 arrangement where they didn't have to.</piece-of-writing>
<piece-type>Essay / opinion piece</piece-type>
<current-pass>Full 5-pass edit</current-pass>
<mode>Mixed — diagnostic for structure + paragraph passes, surgical for sentence + word passes</mode>
<publication-target>The Atlantic or similar — longform digital with strong voice</publication-target>
<what-you-want-preserved>The contrarian thesis. The direct voice. The specific claim that RTO is about management paradigm, not productivity.</what-you-want-preserved>
<known-concerns>It feels repetitive in places. The middle meanders. I'm not sure if the ending lands.</known-concerns>
🤖 Output

Editing Pass: 'Remote Work Isn't About Location' — opinion essay, ~430 words current

A 5-pass editing protocol runs 5 separate passes (Structure → Paragraph → Sentence → Word → Proofread) each with one purpose, producing publication-grade prose through layered discipline. Based on workflows from editors at top literary magazines, writers who run mixed-purpose edits waste 40-60% of their revision time and produce worse results than writers who run single-purpose passes in order. Your draft has a strong thesis, direct voice, and a genuinely contrarian frame — but it suffers from structural repetition (you state the thesis 3 times), paragraph-level density (one long block instead of 4-5 paragraphs), sentence-level rhythm issues, and word-level slackness. This 5-pass edit diagnoses structure + paragraph issues, then surgically rewrites at sentence + word level, producing a publication-ready version at ~550 words with tighter argument and stronger ending.

The 5-Pass System (Context)

  • Pass 1: STRUCTURE. Does the argument work? Does the piece go where it needs to?
  • Pass 2: PARAGRAPH. Does each paragraph earn its place? Are they the right size?
  • Pass 3: SENTENCE. Does each sentence say its one thing? Is rhythm working?
  • Pass 4: WORD. Is every word the right word? Cut the slack.
  • Pass 5: PROOFREAD. Zero errors.

Pass 1: STRUCTURE (Diagnostic)

Your current structure:

1. Hook: 'It's not about location' (1 sentence)

2. Set up: both sides arguing wrong thing (2 sentences)

3. Thesis: remote work exposed visibility-based management (2 sentences)

4. Expansion: middle management was premised on visual presence (3 sentences)

5. Repeated thesis, slightly reframed: RTO is about management class restoration

6. Why debate is charged: management paradigm shift

7. Ending question: what are we really debating?

Structural diagnosis:

Issue 1: Thesis stated 3 times (paragraphs 3, 5, 7). Once is punch. Twice is reinforcement. Three times is padding. Cut to 2 statements max.

Issue 2: No clear movement. The piece states the thesis, then states it again, then states it again. Readers want ADVANCEMENT — each paragraph should take the argument one step further. Yours loops.

Issue 3: Missing concrete example. The whole essay is abstract. You make a structural claim about middle management but never show a specific manager, a specific company, a specific moment. Abstract arguments in opinion pieces need 1-2 concrete anchors.

Issue 4: Ending is a question, not a landing. 'So what are we really debating...?' is a rhetorical move that works — but only if the answer is fresh. Yours repeats the thesis once more. Consider a stronger landing image or claim.

Issue 5: Length is awkward. At 430 words, the piece is too short for an Atlantic-style longform essay (usually 1,500-3,000) and too long for a pure op-ed punch (usually 600-900). Decide: tighten to 600-700 (op-ed) OR expand to 1,800+ with example, research, opposition (longform).

Recommended structural fix:

  • Keep the contrarian frame
  • Add ONE concrete example (a specific RTO policy, a specific manager-type, a specific moment)
  • State thesis clearly twice (once in hook, once in landing) — not three times
  • Replace middle-paragraph repetition with ADVANCEMENT (each paragraph takes argument forward)
  • Land with a strong claim, not a question

Pass 2: PARAGRAPH (Diagnostic)

Current state: The whole piece is ONE PARAGRAPH. 430 words in one block. This is a serious readability issue for a longform essay target.

Paragraph breakdown suggested (with new structure):

  • Paragraph 1 (Hook): 'It's not about location' — 60-80 words
  • Paragraph 2 (Both sides wrong): The shared premise is flawed — 80-100 words
  • Paragraph 3 (Concrete example): A specific manager, a specific moment — 120-150 words
  • Paragraph 4 (The real thesis): Remote work exposed visibility-based management — 100-120 words
  • Paragraph 5 (Why it's existential): Middle management without visual verification — 120-150 words
  • Paragraph 6 (The emotional charge): Why RTO is fought so hard — 80-100 words
  • Paragraph 7 (Landing): What we're actually debating — 80-100 words

Total: ~650-800 words. The piece doubles to proper longform length while getting TIGHTER argumentatively because each paragraph does specific work.

Pass 3: SENTENCE (Surgical — rewritten)

I'll rewrite the piece at sentence level, preserving your thesis and voice. Here's the sentence-pass output:

---

The thing about remote work that I've been trying to articulate for three years is that it isn't actually about location.

Everyone keeps saying it is. CEOs say we need to be in the office for collaboration, culture, innovation, the magic of the hallway encounter. Employees say we want to work from home because commutes are dead time and pants are an unnecessary tax. Both sides are arguing about the wrong thing.

What remote work actually changed was visibility. For the first time in the history of knowledge work, it became impossible to confuse being in the building with doing the job. The proxy that had quietly substituted for management — 'I can see you at your desk, therefore you must be working' — stopped functioning.

This turned out to be an existential problem for a specific kind of manager. Not all managers. A specific kind: the ones who built their authority on visual verification of presence.

A lot of middle management, across knowledge-work industries, was premised on the ability to stand in a doorway and see people hunched over keyboards. When that was taken away — suddenly, in March 2020 — those managers discovered they had nothing. They didn't know what their teams actually did. They didn't know what output looked like. They didn't know how to evaluate performance without presence as a proxy.

So the return-to-office push isn't, fundamentally, about productivity. The research on productivity is ambiguous at best. It's about the restoration of a management class that cannot function without the ability to see workers in chairs.

This is also why the debate is so emotionally charged. It's not just logistics. It's whether we're going to continue running companies on a management paradigm that no longer makes sense — or whether we're going to force ourselves to build something new. The people who need the old paradigm to continue have enormous incentive to force us back. The people who have moved past it don't need the office.

So what are we really debating when we debate return-to-office? Not location. Not productivity. Whether a generation of managers is finally going to have to learn to manage — or whether they're going to successfully drag everyone else back to a pre-2020 arrangement where they didn't have to.

---

What changed at sentence level:

  • Broke into paragraphs (Pass 2 concern addressed)
  • Tightened opening repetitions ('everyone keeps saying it is' → combined with CEO/employee framing)
  • Added concreteness ('stand in a doorway and see people hunched over keyboards')
  • Added a specific moment ('suddenly, in March 2020')
  • Removed one of the thesis repetitions
  • Sharpened sentence rhythms (short → medium → long variation)
  • Strengthened ending — kept the question but made the answer punchier

Pass 4: WORD (Surgical)

At word level, tightening further:

Examples of word-level cuts:

  • 'CEOs say we need to be in the office for collaboration, culture, innovation, the magic of the hallway encounter.' — Strong list. Keep.
  • 'Employees say we want to work from home because commutes are dead time and pants are an unnecessary tax.' — 'Pants' adds voice and humor. Keep.
  • 'became impossible to confuse' — could tighten to 'could no longer confuse' for slight punch.
  • 'the ones who built their authority on visual verification of presence' — tighten to 'the ones whose authority came from seeing people at their desks.'
  • 'so the return-to-office push isn't, fundamentally, about productivity' — 'fundamentally' is slack. Cut: 'so the return-to-office push isn't about productivity.'
  • 'The research on productivity is ambiguous at best.' — Keep. Concrete addition.
  • 'whether a generation of managers is finally going to have to learn to manage' — clean, punchy. Keep.

Final word-pass version sits at ~560 words. Further cutting to 450 is possible but would cost voice/texture. At Atlantic length-target, this is appropriate.

Pass 5: PROOFREAD

Checking for:

  • Typos: none detected
  • Punctuation: clean
  • Consistency (numbers, tense, em-dash style): clean
  • Spelling (British vs. American): American English throughout, consistent

Proof clean.

Before/After Examples

Example 1: Opening

  • Before: 'The thing about remote work that I've been trying to articulate for three years is that it's not actually about location. Everyone keeps saying it is. CEOs keep saying we need to be in the office. Employees keep saying we want to work from home. Both sides are arguing about the wrong thing.'
  • After: [First two paragraphs of rewrite — breaks into 2 paragraphs, adds specific CEO/employee content]

Example 2: Middle claim

  • Before: 'It turns out — and this is what everyone is slowly figuring out — that a lot of middle management, across knowledge-work industries, was premised on the ability to visually verify that people were present.'
  • After: 'A lot of middle management, across knowledge-work industries, was premised on the ability to stand in a doorway and see people hunched over keyboards.'
  • Change: Cut the 'It turns out' meta-comment (slack). Replaced 'visually verify that people were present' (abstract) with 'stand in a doorway and see people hunched over keyboards' (concrete image).

Example 3: Ending

  • Before: 'So what are we really debating when we debate return-to-office? Not location. Not productivity. We're debating whether a whole generation of managers is going to have to finally learn how to actually manage — or whether they're going to successfully drag everyone else back to the pre-2020 arrangement where they didn't have to.'
  • After: 'So what are we really debating when we debate return-to-office? Not location. Not productivity. Whether a generation of managers is finally going to have to learn to manage — or whether they're going to successfully drag everyone else back to a pre-2020 arrangement where they didn't have to.'
  • Change: Cut 'We're debating' (redundant with the question). Small sharpening.

What I Did NOT Touch

✓ Your contrarian thesis — preserved exactly

✓ Your direct voice — preserved; if anything, tightened

✓ The specific claim that RTO is about management paradigm, not productivity

✓ Your voice-markers (wry asides, direct address, specific metaphors)

✓ The rhetorical arc (hook → thesis → mechanism → stakes → landing)

What's Left For Later Passes

Not done here — worth considering in a second round:

  • Adding a concrete example. Structure pass flagged this. Essay would strengthen with 1 specific RTO policy (Amazon's 5-day mandate? Disney? Specific manager-type?).
  • Research citation. You claim 'research on productivity is ambiguous.' Citing 1-2 specific studies would harden the claim.
  • Extending to 1,800-2,500 words for true longform. Currently 560. If Atlantic longform, expand middle with example + research + counter-position.

These are DEVELOPMENTAL concerns, not editing concerns. Noted but not addressed in 5-pass edit.

Readiness for Next Pass

Current state: This version is ready for one more targeted pass at paragraph level (if you want to add the concrete example + research). Otherwise publishable as-is for shorter-format outlets.

Key Takeaways

  • 5-pass editing, IN ORDER. Structure → Paragraph → Sentence → Word → Proofread. Don't mix passes.
  • Your piece's biggest issues are structural (thesis stated 3x, no paragraph breaks, no concrete anchor) — those need to be fixed FIRST before sentence polishing.
  • The surgical sentence-level rewrite tightens from 430 → 560 words while adding structure, concreteness, and rhythm variation. Paradoxically, longer here means tighter argument.
  • For Atlantic-level longform target: expand middle with one concrete example + 1-2 research citations. Currently at op-ed length, not longform length.
  • The thesis, voice, and contrarian frame are preserved exactly. Only connective tissue was touched.

Common use cases

  • Essayists revising drafts before submission to literary magazines
  • Bloggers/newsletter writers polishing posts before publication
  • Novelists doing self-edit passes before sending to agent/editor
  • Copywriters revising high-stakes client work
  • Business writers polishing reports, memos, and executive communications
  • Academics revising papers (especially combined with academic clarity surgeon)
  • MFA students learning professional editing discipline
  • Writers preparing submissions for pitches, grants, or job applications
  • Freelance editors running systematic client edits

Best AI model for this

Claude Opus 4 or Sonnet 4.5. Multi-pass editing requires different analytical modes — structural, sentence-level, word-level — each needing careful attention. Top-tier reasoning matters.

Pro tips

  • Run the passes IN ORDER. Fixing word-level problems in a paragraph that will get deleted in the structure pass is wasted effort. Top-down is non-negotiable.
  • Each pass is ONE pass. Structure pass = only structure. Paragraph pass = only paragraph. If you find yourself fixing word choice during the structure pass, note it and keep moving. Single-purpose passes are 3-5x faster than mixed passes.
  • Between passes, WAIT 24 hours minimum. Your brain needs to reset between analytical modes. Back-to-back passes produce worse edits — your structural eye bleeds into your sentence eye.
  • Read aloud during sentence pass. Your ear catches what your eye misses. Clunky rhythm, repeated words, awkward transitions — all surface when read aloud.
  • Structure pass asks 'does this ARGUMENT work?' Paragraph pass asks 'does this PARAGRAPH earn its place?' Sentence pass asks 'does this SENTENCE say its one thing?' Word pass asks 'is every WORD the right word?' Keep the questions distinct.
  • Proofread pass is LAST — and most people run it first, which is why they miss errors later introduced during heavier edits. Errors introduced in a word-pass are caught by a final proofread; errors caught in first proofread get re-introduced when you change paragraphs later.
  • Budget your time: Structure pass = 30% of edit time. Paragraph = 25%. Sentence = 25%. Word = 15%. Proofread = 5%. Most writers spend 80% on words and 20% on structure — the inverse of what works.
  • If you're running the passes on someone else's work (you're the editor), get explicit agreement on what 'editing' they want. Line edit ≠ developmental edit ≠ copy edit. Be specific about which pass layers you're running.

Customization tips

  • If you're editing your own work, wait 48-72 hours between drafting and first edit pass. Fresh eyes catch what exhausted eyes miss. Editing your own writing the same day you wrote it is the worst time to do it.
  • For short work (under 1,500 words), collapse the 5 passes into 3: structural + paragraph combined, sentence + word combined, proofread separate. Full 5-pass is overkill for blog posts.
  • Print your piece for the sentence pass. Reading on-screen changes what you catch. Writers read ~25% more critically on paper than on screen — empirically validated.
  • Keep a 'deferred fixes' list during each pass. When you notice a word-level issue during the structure pass, write it down but don't fix it. You'll catch it in the word pass, and your brain will stay in structure-mode longer.
  • After all 5 passes, have someone else read it (friend, writing group, paid editor). You can only self-edit to ~85% quality. The last 15% requires other eyes.

Variants

Blog Post Mode

For 1,500-3,000 word posts. Collapses the 5 passes into 3 pragmatic passes for shorter work.

Novel Chapter Mode

For book chapters (3,000-8,000 words). Adds scene-level and dialogue-level considerations within the 5 passes.

Business Document Mode

For memos, reports, executive communications. Emphasizes structure and paragraph passes; relaxes some stylistic word-pass norms.

Academic Paper Mode

For journal articles. Combines with academic clarity surgeon — adds citation-accuracy and methods-preservation considerations.

Frequently asked questions

How do I use the Editing Pass Protocol — The 5-Pass System For Publishable Prose prompt?

Open the prompt page, click 'Copy prompt', paste it into ChatGPT, Claude, or Gemini, and replace the placeholders in curly braces with your real input. The prompt is also launchable directly in each model with one click.

Which AI model works best with Editing Pass Protocol — The 5-Pass System For Publishable Prose?

Claude Opus 4 or Sonnet 4.5. Multi-pass editing requires different analytical modes — structural, sentence-level, word-level — each needing careful attention. Top-tier reasoning matters.

Can I customize the Editing Pass Protocol — The 5-Pass System For Publishable Prose prompt for my use case?

Yes — every Promptolis Original is designed to be customized. Key levers: Run the passes IN ORDER. Fixing word-level problems in a paragraph that will get deleted in the structure pass is wasted effort. Top-down is non-negotiable.; Each pass is ONE pass. Structure pass = only structure. Paragraph pass = only paragraph. If you find yourself fixing word choice during the structure pass, note it and keep moving. Single-purpose passes are 3-5x faster than mixed passes.

Explore more Originals

Hand-crafted 2026-grade prompts that actually change how you work.

← All Promptolis Originals