Scoring criteria
See "Overall 7.3" and you'll want to know — "how is 7.3 different from 8.1? what's an 8?" This article explains.
What 0–10 means
| Range | Meaning |
|---|---|
| 9–10 | Rare — close to submission-ready |
| 7–9 | Strong, with a few patches |
| 5–7 | Mid, several areas to fix |
| 3–5 | Structural issues; major revision |
| 0–3 | Very early draft |
Most first drafts land 5–7 — that's normal, not failure.
How scores are computed
Each dimension has 4–6 sub-criteria; AI scores each 0–10, weighted average yields the dimension score.
Example: "Structure" (5 subs)
- Act break clarity (25%)
- Turning point impact (25%)
- Chapter ordering coherence (15%)
- Opening setup (20%)
- Ending completeness (15%)
Each sub has its own rubric.
Full rubrics are in the report's "Detailed scoring criteria" expandable section.
Why scores are directional, not absolute
A few reasons:
1 · LLM nature
Different model versions / temperatures can shift the same book by 0.5.
2 · Persona choice
Strict personas vs generous personas → same book, different scores.
3 · Your category
A "7" in literary fiction vs "7" in popular fiction don't mean the same thing.
Comparing same-book runs is what's meaningful
Revise and rerun is more useful than any single absolute score:
- Run 1: 6.2
- Run 2 (after revision): 7.5
- → your revisions are working
See: Reading history & revisit
Alignment with external reviews
If you also have a human beta reader score — you'll find their scores usually align directionally with Slima's (not always with the same number).
E.g. human reader gives 8, Slima gives 7.5. Direction matches is the signal; absolute numbers aren't.
Related
Was this helpful?