Getting the Most from Writing Analysis
What to submit
Red Ink works best on passages of 500-5000 words. Too short and there isn't enough material for meaningful analysis. Too long and the feedback becomes general rather than specific.
Submit self-contained sections — a complete chapter, a full essay, a finished blog post. Fragments without context produce less useful feedback because the AI can't evaluate structure or pacing without seeing the whole arc.
How to read feedback
Read the overall assessment first to calibrate. Then focus on specific dimension scores. The most valuable feedback is usually in dimensions you weren't thinking about — if you focused on dialogue during revision but Red Ink flags pacing, that's the insight you need.
Don't weight all dimensions equally. For a fast-paced thriller, pacing matters more than description. For literary fiction, voice and description may matter more than pacing. Know your genre and prioritize accordingly.
When to disagree
AI analysis is not always right. Specific cases where you should trust your own judgment:
Intentional choices. If you deliberately wrote a long, slow opening to create a specific mood and the AI flags it as "slow pacing," that might be the AI missing your intent. Stand by intentional craft decisions.
Genre conventions. AI models may not fully understand genre expectations. A romance novel's pacing feels different from a thriller's, and that's correct, not a flaw.
Voice. If the AI suggests your voice is "inconsistent" but you're writing a first-person narrator with a deliberately inconsistent personality, the AI may be misreading character voice as authorial weakness.
The rule: understand why the AI flagged something before deciding whether to agree. If you can articulate why you made that choice, keep it. If you can't, the AI might be right.
Building a revision workflow
- Write the first draft without analysis
- Self-revise once (your own judgment first)
- Submit to Diagnostic — get the broad strokes
- Revise based on Diagnostic feedback you agree with
- Submit to Diagnostic again — verify improvements
- When satisfied, submit to Jury for final assessment
- Use Jury consensus for the final polish
This iterative approach uses AI as a feedback loop, not a replacement for your own editorial judgment.