Back to Docs
Core Concept
Scoring Methodology
Scoring helps teams summarize a message review, but the score is only useful when read alongside the underlying objections and confusion points.
Key points
- Clarity as a directional summary
- Confidence tied to run depth
- Objection review alongside score
- Trend and comparison support
How to read the signal
- Start with clarity and decision confidence, then inspect the qualitative evidence.
- Large score moves matter less if the underlying objections stay the same.
- Use comparisons across revisions to decide whether the rewrite actually improved the message.
What the score is not
- It is not a revenue guarantee or a substitute for live experiment data.
- It is not independent from persona quality or scenario framing.
- It should not be used alone for executive approval when the stakes are high.
In plain terms
- Robust mode evaluates the same message five times. Its 95% confidence interval summarizes variation in those generated evaluations; it does not describe the real buyer population or promise that a result will occur 95 times out of 100.
- A narrow interval means the five generated scores were close under this setup. A wide interval means they varied; inspect the objections and inputs instead of assigning one cause automatically.
- Treat a high score with a wide interval more cautiously than one with a narrow interval. The interval is a stability aid for this synthetic review, not standalone market evidence.