Scoring model
01How open potential is calculated
- Base score starts mid-band; good findings add less than bad findings subtract (strict calibration).
- Excellent is rare (≥90). Good / fair / needs-work bands are intentionally strict.
- Variants are ranked live; near-duplicates are flagged so A/B tests stay meaningful.
- Optional preview text is scored for pairing (truncation + support of the subject promise).
Pillars
02What gets scored
- Length / mobile truncation risk
- Framework detection (how-to, question, reply-style, etc.)
- Power words and curiosity structure
- Spam lexicon dual score (content risk vs open potential)
- Formatting noise (caps, punctuation, emoji density)
- Preheader pairing when provided
Limits
03What this does not mean
- Not a forecast of your list’s open rate (list quality, sender, timing dominate).
- Not a guarantee any ESP filter will treat the line the same way.
- Does not score body HTML (use the Spam Score Checker for body risk).