Getting Feedback
Run a beta read session and interpret the results.
Last updated March 2026
Overview
A beta read session is where you submit your text to one or more beta reader personas and receive detailed, perspective-driven feedback. Each reader analyzes your work independently and delivers their critique through their unique voice and area of expertise.
Running a beta read requires a Writer+ plan or higher.
Starting a Session
To start a beta read session:
- Open the Tools panel (right sidebar) in the writer interface.
- Click the Beta Read tab or button.
- Select 1–5 beta reader personas for your session.
- Optionally provide context about your project (see below).
- Click Start Beta Read to begin the analysis.
The beta readers will analyze your current draft content. Make sure you have enough text in your draft before starting — the minimum is 300 characters (roughly 2–3 sentences). The more text you provide, the more detailed the feedback.
Choosing Your Readers
You can select between 1 and 5 beta readers per session. More readers means more perspectives, but also higher token cost.
Choose readers strategically based on what you need:
- First draft? Start with The Architect (plot structure) and The Page-Turner (engagement). Big-picture feedback first.
- Revising? Add The Hawk (prose craft) and The Empath (emotional resonance) for line-level feedback.
- Worldbuilding-heavy scene? Include The Cartographer (world consistency) to catch setting issues.
- Genre fiction? Add a Genre-Specific reader (built-in or from the marketplace) to check genre conventions.
Providing Optional Context
Before starting the session, you can provide additional context to help readers give more informed feedback. This context is optional but can significantly improve the quality of analysis:
- Project title — the name of your novel or story.
- Genre — helps readers calibrate expectations (a slow burn in literary fiction is pacing; in a thriller, it might be a problem).
- Tone — the intended mood and voice of the piece.
- Themes — major themes the readers should be aware of.
- Logline — a one-sentence summary of the story.
- Story Bible context — additional background information, up to 3,000 characters. This can include character descriptions, world rules, or narrative goals for the section being analyzed.
- Chapter title — the name of the current chapter or scene.
How the Analysis Works
When you start a session, your selected readers begin analyzing your text in parallel. This means all readers start working at the same time, and results appear as each reader finishes — the fastest readers show their feedback first while others are still reading.
You'll see real-time progress as results stream in:
- Persona started — the reader has begun analyzing your text.
- Persona complete — the reader's feedback is ready to view.
Results are progressive. You can start reading feedback from the first reader to finish while waiting for the others. There's no need to wait for all readers to complete before reviewing.
Reading the Results
Each reader delivers feedback structured around their specialty. Depending on the reader and your text, feedback typically covers:
- Pacing analysis — where the text flows well and where it stalls.
- Character consistency — whether characters behave authentically.
- Plot logic — cause-and-effect chains, setup/payoff, holes.
- Prose quality — sentence-level craft observations.
- Engagement assessment — where a reader would keep going or put the book down.
Each reader's feedback is independent. Compare their perspectives to find consensus. When The Architect and The Page-Turner both flag the same scene as problematic, that's a strong signal. When only one reader raises an issue, weigh it against your own instincts.
Input Requirements
Beta read sessions have specific input constraints:
| Constraint | Value |
|---|---|
| Minimum text length | 300 characters |
| Maximum text length | 80,000 characters |
| Minimum readers | 1 |
| Maximum readers | 5 |
For best results, submit a complete scene or chapter rather than a fragment. Readers give more useful feedback when they can see the full arc of what you're trying to achieve.
Model Override
By default, beta reads use your currently selected AI model. However, you can override the model for the session if you want to use a different one.
Premium models like Claude 4.6 Sonnet and GPT-5.2 tend to produce more nuanced and detailed feedback. Budget models are faster and cheaper but may give less detailed analysis. See Choosing a Model for guidance on model tradeoffs.
Token Cost
Token cost for a beta read session depends on two factors:
- Text length — longer texts require more input tokens per reader.
- Number of readers — each reader is a separate AI call, so 5 readers costs roughly 5x what 1 reader costs.
All costs come from your Genesis Token balance. The model you choose (or that's selected by default) affects the per-token rate. Plan accordingly if you're running full 5-reader sessions on long chapters.
Tips for Better Feedback
- Submit complete scenes. A full scene with a beginning, middle, and end gives readers enough material to assess pacing, structure, and emotional arc.
- Provide genre context. Readers calibrate feedback based on genre expectations. What works in literary fiction may not work in a thriller.
- Run sessions at different stages. Get structural feedback on early drafts, then prose-level feedback on later revisions. Don't polish sentences that might get cut.
- Compare across readers. The real value is in the overlap. Issues that multiple readers flag independently are almost certainly worth addressing.
- Re-run after revisions. Made changes based on feedback? Run another session to see if the fixes landed. Different readers may notice new things in the revised version.
- Try community readers. The marketplace has specialized readers for specific genres and writing goals that may give more targeted feedback than the built-in personas.