Manuscript Analysis
How a first-read report turns uncertainty into revision priorities
In short
A thirty-page editorial report that shows what is working, what needs attention, and where the next revision should start.
- Ten-axis editorial review (structure, voice, market fit, audience match, register, and more) with concrete revision steps.
- Personalised to your imprint when a Publisher Overview is configured — otherwise reviewed against general editorial criteria.
- A manuscript synopsis, chapter-by-chapter summaries, an insightful analysis of characters and narrative structure, a draft back-cover blurb, and a publishability assessment.
- Exportable PDF — shareable with authors, acquisition teams, or your editorial board.
You upload a Word file, click a button, and a few minutes later you have a thirty-page editorial report with a chapter-by-chapter breakdown, revision priorities, and even a draft back-cover blurb. If you've ever wondered what's going on under the surface during those few minutes — this is the walkthrough.
When you're faced with a growing pile of unread submissions, a quick analysis helps you cut through the noise. It tells you immediately whether a manuscript's structural potential and estimated editorial effort make it worth your investment — even when the writing still needs significant work.
Your manuscript is scanned automatically on upload
The process starts the moment you upload a .docx file. Before you do anything else, EditBook.ai runs a quick scan in the background. This extracts the basics: the working title, the chapter structure, a list of characters (or core concepts, for non-fiction), word counts per chapter, and the language of the manuscript.
None of this requires you to press a button — it happens automatically. By the time you open the project page, the chapter list is already there.
The quick scan also determines whether the manuscript is fiction or non-fiction. That distinction forms the basis for the rest of the analysis, which adapts its structure accordingly. In fiction, the focus is on character development and how scenes connect; in non-fiction, on the structure of the argument and the quality of its supporting evidence.
The report adapts to your imprint and audience
Before you run the analysis, there's one thing worth knowing: EditBook.ai uses your publisher profile to personalise the report. If you've set up a Publisher Overview in your organisation settings -- describing your imprint, your catalogue, your target audience, and your editorial identity -- the analysis will evaluate the manuscript against that context.
This means the report doesn't just say whether a manuscript is "good" in the abstract. It tells you whether it fits your list, whether the audience matches your readers, and whether the editorial effort is realistic for your team. The Publisher Overview acts as a personalised lens for measures like Publisher Identity Fit and Market Fit.
If no Publisher Overview is configured, the analysis still works -- it simply evaluates the manuscript on general editorial criteria rather than publisher-specific ones.
Choose your reports and AI model
When you're ready, open the analysis panel and click Manuscript analysis. A dialog appears where you select which report components to include and which AI model to use.
You always get the main analysis report. Optionally, you can add an Author Feedback report — a separate document written in a constructive, author-facing tone that you can send directly to the writer without having to rewrite your internal notes.
Then you pick a model. Each option shows a quality rating and an estimated credit cost. A faster model like Google's Gemini 3.0 Flash uses less than one credit, while a more thorough model like Claude Opus costs more but tends to produce sharper observations on style, character, and structure.
Click the button, and the analysis begins. The system first computes quantitative readability metrics — Flesch Reading Ease, lexical density, vocabulary diversity, sentence length distributions — before any AI model is involved. Then the AI reads the full manuscript, generates the editorial report, produces scores, and writes the optional feedback document. The entire process typically takes a few minutes.
Ten editorial signals, a full report, and concrete next steps
I. Ten signals show strengths and improvement areas at a glance
When the analysis finishes, you see a summary card with the overall verdict, star ratings, and ten scores displayed as progress bars with one-line explanations. This is designed as a quick overview — you can see at a glance where the manuscript is strong and where it needs attention.
Click on any score to expand it. Each score includes a short explanation, a longer rationale with evidence from the text, and a concrete improvement action you can take. This makes the scores directly actionable — you don't just see a number, you see what to do about it.
The signals are calibrated against explicit editorial anchors:
| Range | Meaning |
|---|---|
| 0–2 | Needs a substantial rethink |
| 3–4 | Needs significant development |
| 5–6 | Mixed / conditionally viable |
| 7–8 | Strong / solidly publishable |
| 9–10 | Exceptional |
The ten metrics are: strategic publishability, publisher identity fit, market fit and sales potential, originality and USP, structure and pacing, character or argument strength, impact and engagement, writing style and voice, readability and clarity, and technical correctness.
Spelling and grammar errors are explicitly not penalized in the review — the system knows that AI-assisted editing can fix those efficiently. What matters more is the kind of issue that needs an author or editor's judgment: plot holes, structural drift, loss of voice, or underdeveloped evidence.
II. Readability: statistical metrics plus AI reading-level assessment
The readability report combines statistical metrics with an AI-assessed reading level. It shows sentence and word length distributions, vocabulary profile, Flesch and other readability scores, and arrives at a CEFR level (A1 through C2) with an estimated reading age.
This matters for books targeting younger audiences or specific reading levels — a book can score "easy" on a formula like Flesch because it uses short sentences, while the themes and vocabulary actually demand an older, more experienced reader. The model is instructed to correct for this.
III. A 4,000-word editorial report in six sections
The core of the analysis is a structured editorial report of approximately 4,000 words, organized into six major sections:
- Summary — the most important points from the entire report.
- A. Summary — plot synopsis (or thesis summary for non-fiction), theme, genre and audience, character overview (or concept overview), and chapter-by-chapter summaries.
- B. Content Analysis — nine subsections covering target audience and market fit, publisher identity and portfolio fit (including a "gatekeeper analysis" with a brand alignment checklist), thematic consistency and originality, emotional or practical impact, narrative structure and arc, scene connectivity and functional flow, logical consistency, character analysis (or argument strength), and a minimum of three concrete improvement proposals. Each section states clearly where the manuscript excels and what would make it stronger.
- C. Writing Style Analysis — ten subsections covering contemporaneity, language level and readability, narrative perspective and tense consistency (with at least three concrete examples), voice and tone, show-don't-tell and imagery, dialogue quality (or anecdote quality for non-fiction), rhythm and stylistic devices, characteristic passages that show what is working and what can be improved, length and pacing metrics, and technical correction investment with error density estimates and copyediting hour calculations.
- D. Publication Strategy — illustration recommendations, a title analysis with at least ten alternative suggestions, cover design proposals, a back-cover blurb of 120–140 words, and promotion ideas.
- E. Publishing Advice — an overall verdict, an analysis matrix (manuscript status, sales potential, editorial effort, author capacity, publisher fit), must-fix action items, nice-to-haves, and estimated editorial turnaround times per role.
The report is written in the language of your choice — it doesn't have to match the language of the manuscript.
IV. A ready-to-send feedback document for the author
If you selected this option, the system generates a second document: a feedback report written directly to the author. It covers the same ground — structure, characters, style, readability — but in a constructive, actionable tone. It includes rewrite examples, concrete suggestions per section, and a careful assessment of audience fit. This is the document you can send to the writer as-is.
A foundation for decisions -- not a replacement for judgment
The analysis is designed to answer the question a publisher asks when a manuscript arrives: Is this worth the time it would take to find out? It provides the structured overview that makes an informed decision possible — across content, style, market fit, and production effort — without requiring a full read first.
What the analysis is not is a replacement for editorial judgment. It won't tell you whether a book moved you, whether the author's voice is one you want to champion, or whether the market is ready for this particular story. Those remain editorial decisions. But it gives you a consistent, thorough foundation to make them from — regardless of whether the manuscript arrived on a Monday morning or buried in a Friday afternoon pile.
The analysis module produces up to four documents per manuscript: the main analysis report, author feedback, readability assessment, and a structured scoring breakdown. All are available as downloadable Word files from the project page.
Ready to try it on a manuscript that needs attention?
See how structured analysis and track-changed suggestions support the editor's judgment.