Methodology
How we score every book
Every listing on the ProseCreator marketplace is analyzed by our Digital DNA pipeline — 12 craft dimensions, 5 quality tiers, and transparent AI disclosure labels. No paid placements. No black-box scoring. Methodology lives here so buyers can decide for themselves.
The 12 dimensions
Each book is scored on the dimensions below. Scores roll up into a composite DNA score (0–100) and a percentile-within-genre.
Voice
Narrative voice consistency, point of view, and authorial signature.
Pacing
Scene-level rhythm, beat distribution, and reader retention pressure.
Dialogue
Character-distinguishable speech, subtext density, and information transfer.
Structure
Macro plot architecture against reference frameworks (3-act, hero’s journey, etc).
Character Arc
Want/need progression, transformation depth, and consistency of motivation.
World-Building
Internal consistency, sensory specificity, and lore depth without exposition dumps.
Theme
Thematic clarity, weave through plot, and resonance with stated genre conventions.
Prose Craft
Sentence-level rhythm, image density, and avoidance of filler / hedging.
Originality
Style-fingerprint distance from the genre median (high = distinctive).
Continuity
Internal consistency across chapters — facts, timelines, character details.
Emotional Range
Variance in scene tone and emotional payload across the manuscript.
Genre Fit
Adherence to genre expectations measured against curated reference sets.
Genre weighting
Different genres prize different craft signals. A literary novel is weighted differently from a cozy mystery. Our percentile rankings compare a book against its genre cohort, never a global average. We publish each genre's weight matrix so authors can audit it.
- Literary fiction — Prose Craft 1.4×, Theme 1.3×, Voice 1.2×.
- Thriller / suspense — Pacing 1.4×, Structure 1.2×, Continuity 1.2×.
- Romance — Character Arc 1.3×, Dialogue 1.3×, Emotional Range 1.2×.
- SFF — World-Building 1.4×, Originality 1.3×, Theme 1.2×.
Quality badges
Five-tier discovery hierarchy. Buyers can filter by tier; authors graduate as the manuscript improves.
- Developing
Pre-quality / unreviewed listings. Available but not DNA-validated.
- Listed
Listed but no DNA score yet — analysis in progress.
- Standard
Passed the quality threshold. Solid craft signals across the board.
- Quality Verified
DNA score >= 80. Top-quartile manuscripts on at least 8 of 12 dimensions.
- Editor's Choice
Top-tier curated picks. Hand-reviewed by ProseCreator editors.
AI transparency labels
Every listing discloses how AI was used during creation. We use a three-tier disclosure scheme. Mislabeling is grounds for delisting.
- Human-written
Fully human-authored. AI tools may have been used as spellcheck or research aids only.
- AI-assisted
AI used as a co-author or editing aid. The human author remains the primary creative voice.
- AI-created
AI is the primary author. Disclosed transparently so buyers can choose.
Evidence verification
Every DNA report cites the passages that drove each score. The Voice score, for example, points to the specific sentences that established voice consistency or revealed inconsistencies. Buyers can audit our reasoning by clicking through to the evidence; authors can dispute scores through the appeal process.
- Cited passages — every dimension links to source text excerpts.
- Reproducible — DNA reports are versioned. Re-runs with the same input produce the same score.
- Appealable — authors can dispute findings through the Author Center.
How to read your report
- Composite DNA score — the headline 0–100 number. Treat anything above 80 as Quality Verified.
- Percentile within genre — “Top 8% in thriller” is more meaningful than the raw composite.
- Per-dimension bars — quickly scan strengths and weaknesses. Bars below 40 indicate the dimensions to focus on.
- Style fingerprint neighbors — books with similar prose signature. Use this to find readers and comp titles.
- Evidence drilldown — click any score to see the cited passages that drove it.