Docs › AI Readers › Scoring criteria
Scoring criteria
Last updated June 10, 2026 · 3 min read
The report is full of scores — this article: where they appear, how to read them, and why they're directional not absolute.
Where scores appear
| Location | Score |
|---|---|
| Reader feedback 5 diagnostics | Each diagnostic's sub-scores, 0-10 (e.g. Opening appeal's Hook / Positioning; Character analysis's Protagonist / Motivation / Dialogue / Voice) |
| Each reader | In "full reader feedback", every reader scores Opening appeal / Character / Pacing / Overall 0-10 |
| Overview | Continue-reading willingness, Recommendation index (0-10); Kindle predicted rating (★, out of 5) |
What 0-10 roughly means
| Range | Rough meaning |
|---|---|
| 8-10 | Strong, near submission-ready |
| 6-8 | Good, with room to improve |
| 4-6 | Mid, several areas to fix |
| < 4 | Major revision needed |
Most first drafts land mid-range — normal, not failure.
Why scores are directional
- LLM nature: different models / settings can shift the same passage by a few tenths.
- Reader choice: strict vs generous readers score differently (the same passage might get 3 from Marcus, 7 from Olivia).
- Your category: a "7" in literary fiction isn't a "7" in popular fiction.
Comparing same-book reruns is what matters
Revise and rerun, then see if the score climbs — far more useful than any single absolute number. See: Reading history & revisit
Related
Was this helpful?